Job search assistance for vocational students

The intervention consists of two parts. The first part is a counseling program which mainly takes place at the class level, but which also contains some individual sessions. The second part takes place after the end of the school year and is an individual job search assistance program.

Part 1: Counseling program

In this counseling program, in each treated class, caseworkers provide general information about the employment prospects of the students, the students learn about the relevance of specific skills on the labor market, and they apply tools to define and to measure their own skills. Moreover, they learn how to search for a job, how to apply to a job, and how to present themselves in a job interview. The students will also meet employers in their local labor market and visit firms. These collective sessions are combined with individual sessions. In these individual sessions, the students discuss the professional project and their job search with the caseworker. They also have access to a psychologist who advises them in developing their professional project. Overall, this program consists of around 25 hours; around 20 hours for the collective sessions and around 5 hours for the individual sessions. This first treatment takes place from January 2022 to June 2022.

Part 2: Individual job search assistance program

In this part, the treated students get an intensified support with the possibility of mobilizing the services offered by the PES. This includes the proposal of job offers to the students and the recommendation of the students to potential employers. This might also include the placement of internships for the students. In this case, the caseworker would continue supporting the students during the internship in their job search process. The goal of this part of the program is to help the students to find a stable job. This second treatment will take place from mid-June 2022 until October 2022.

The intervention will be repeated in 2023 with the same schools selected for the intervention in 2022 (the treatment status will not change over time) and with additionally selected schools.

2022-01-16

2023-10-15

(1) Employment probability; (2) Earnings; (3) Enrollment in (higher) education.

We rely on administrative registers which provide rich information about all students of vocational schools and their situation after school. They contain information about employment in public and private sectors, earnings, participation in training programs and enrollment in the educational system.

Based on these administrative data, we observe for every individual in our sample the employment status after exiting vocational school, with or without diploma. Depending on the availability of sources, due to confidentiality issue, individuals can be tracked, without limitation in time, either every 6 months or on a daily basis. We will estimate the probability of being employed at time t after leaving the vocational school. Besides the employment state at a specific point in time, we will also be able to observe the earnings of employed individuals. This allows us to construct yearly earnings for each individual in our experiment and we will be able to estimate the impact of the intervention on earnings in a specific year as well as on cumulated earnings over several years.

We will use the administrative data on the enrollment in the educational system to measure the impact of the interventions on the probability of entering educational tracks which lead to a higher educational degree or to another educational degree of similar level. We will additionally be able to observe whether they finish these educational programs.

We will explore effect heterogeneity of the interventions with respect to the situation on the local labor market. For this we will use regional data on sector- and occupation-specific unemployment rates as well as information on the employment rates of previous cohorts of vocational students who graduated in the same vocational schools.

(1) Sector of Employment; (2) Occupation of this job; (3) Local Job creation; (4) Final grades.

Based on administrated data, we will be able to observe the sector of employment and the type of occupation the individuals enter. Moreover, we will be able to investigate whether there are additional jobs created in areas with treated schools. For this we will use administrative data on jobs at the local and sectoral level.

Based on administrative records from the educational system, we will be able to analyze the impact of the intervention on the probability of finishing the vocational school with a diploma, and on the final grades.

We will explore effect heterogeneity of the interventions with respect to the situation on the local labor market. For this we will use regional data on sector- and occupation-specific unemployment rates as well as information on the employment rates of previous cohorts of vocational students who graduated in the same vocational schools.

We set up the experiment in 8 regions in France. In these regions, we have around 400 public vocational schools entering our experiment. The group of treated schools will consist of about 200 schools. The treatment in the schools will be delivered by around 50 counsellors employed by the public employment service (Pôle Emploi). Each of these 50 counsellors will be in charge of 4 schools.

Our starting points for the randomization are the local offices of the public employment service in the participating regions. For each local office, we identify the vocational schools which are reachable within less than 60 minutes by car. We select those local offices which have a sufficiently large number of vocational schools in their area (usually those with at least 8 vocational schools reachable within less than 60 minutes by car since each agency is in charge of 4 schools). For each of these schools, we observe the average share of students which have been employed 6 months after finishing the school in the years 2018 and 2019. Within schools we additionally observe this employment rate at the class level. Each class corresponds to one specific occupational degree.

In a first step, we randomly select schools to participate in the experiment. For this, we stratify schools by average employment rates of the graduates of previous cohorts. This stratification takes place within the schools which are allocated to the same public employment office.

In a second step, we randomly choose within each school half of the classes to participate in the experiment. For this, we stratify the classes by the employment rates observed for graduates who have chosen the same occupational track in the same school in previous years. Students in the selected classes will be offered to participate in the counseling program (part 1 of the intervention).

In a third step, we randomly choose half of the students who participate in the first part of the intervention. The selected students will be contacted by the counsellor of the public employment service and they will be offered to participate in an individual job search assistance program (part 2 of the intervention).

Students choose during the final year whether they want to prepare themselves for entering the labor market (professional track) or whether they want to go on with their studies (higher education track). Our intervention is designed for the first group of students (professional track). If possible, we will collect pre- and post-treatment information about the type of track the students have chosen.

We can distinguish four groups of students. First, we observe outcomes for students who are in the professional track in non-treated schools (group A). Second, we observe outcomes of students in the professional track in non-treated classes (group B) within treated schools. Third, we observe outcomes of students who participate in the group-level treatment and who were not assigned to the individual job counseling program (group C). Fourth, we observe outcomes of students who get the group-level treatment and who are offered the individual counseling program (group D).

Our design allows us to analyze different effects of the interventions:
(1) A comparison of groups D and C will allow us to estimate the additional effect of the individual counseling program compared to the group level program only.
(2) A comparison of groups D and B will give us an estimate of the joint effect of the group level treatment and the individual counseling program.
(3) A comparison of groups D and A will give us an estimate of the joint effect of the group level treatment and the individual counseling program without potential displacement and spillover effects within schools.
(4) A comparison of groups C and B will give us an estimate of the impact of the group level treatment on the outcomes.
(5) A comparison of groups C and A will give us an estimate of the impact of the group level treatment on the outcomes without potential displacement and spillover effects within schools.
(6) A comparison of groups A and B will give us insights about potential displacement effects of the treatment.

The feasibility of the comparisons (2) to (6) will depend on the availability of information about the type of track the students have chosen before the intervention has started (professional or higher education track). In case this information is not available, the corresponding comparisons will be based on all students (professional and higher education track).

We will additionally explore whether the displacement effects differ depending on whether the students in the non-treated schools are outside the labor pools of the labor agencies involved in the experiment. For this, we will divide the group A in two subgroups: Group A1 consists of students in non-treated schools outside the labor pools of labor agencies involved in the experiment, and group A2 consists of students in non-treated schools which are within the labor pools of labor agencies participating in the experiment. We will compare groups A1 and A2 with students in group B.

In 2023, we will repeat the experiment with the schools which have participated in 2022 and we will additionally include new schools. This will lead to around 650 public vocational schools entering our experiment in 2023. The group of treated schools will consist of about 325 schools. The treatment status of schools which already participated in 2022 will not change. The treatment in the schools in 2023 will be delivered by around 90 counsellors employed by the public employment service (Pôle Emploi).

Not available

The randomization is done by the research team with the help of a computer program.

1) At the school level
2) Within schools at the class level
3) Within the treated classes at the individual level

Yes

1) Around 400 schools
2) Around 1,300 classes (on average 3.25 per treated school)

In 2023, we repeat the intervention with an increased sample size. The number of clusters in 2023 will be:

1) Around 650 schools
2) Around 2,100 classes (on average 3.25 per treated school)

Around 23,000 pupils in the treated schools, around 23,000 pupils in the non-treated schools (overall, around 46,000 pupils) In 2023, we repeat the intervention with an increased sample. The number of observations in 2023 will be: Around 37,000 pupils in the treated schools, around 37,000 pupils in the non-treated schools (overall, around 74,000 pupils)

1) 200 treated and 200 non-treated schools
2) Around 650 treated and around 650 non-treated classes in the treated schools
3) Around 11,500 students in the treated classes and around 11,500 students in the nontreated classes. We assume that around 50% of these pupils chose the professional track.
4) In the treated classes, we expect that half of the pupils get the group level treatment, and 25% students are additionally offered the individual counseling program. Which means around 5,750 students in group level treatment and 2,875 in individual counseling program (combined with the group level treatment).

In 2023, we will use the same design and increase the sample size in the treatment arms proportionally to the increase in the overall sample size.

For the power calculation, we assume that we can observe for every class which student is enrolling in the professional track, and we assume that the share of students being enrolled in this track is 50%. The power calculations do not consider that the randomization is stratified based on the employment rate at the school and at the class level. 1) For the comparison of pupils in non-treated classes in treated schools with students in non-treated schools, we get a MDE of 4 percentage points for the employment rate, which corresponds to a relative effect of 8.1% (assuming an average employment rate of 50% and an intraclass correlation of 0.05). 2) For the comparison of pupils in treated classes in treated schools with students in non-treated schools, we get a MDE of 4 percentage points on the employment rate, which corresponds to relative effect of 8.1% (assuming an average employment rate of 50% and an intraclass correlation of 0.05). 3) For the comparison of the pupils in treated classes with pupils in non-treated classes, we get a MDE of 3.1 percentage points for the employment probability, which corresponds to a relative increase of 6.2% (assuming an average employment rate of 50% and an intraclass correlation of 0.05). This MDE corresponds to the average effect of the group-level treatment with and without the individual counselling. 4) For the comparison of the group-level treatment with the group-level treatment combined with the individual counselling we get an MDE of 3.7 percentage points for the employment probability, which corresponds to a relative increase of 7.3% (assuming an average employment rate of 50%).