The second part of the study aims to identify the causal effect of the introduction of CPGE or STS classes on pupils’ career choices after the baccalaureate. The empirical strategy used in the study exploits the temporal and spatial variation in the opening of these programmes between 2007 and 2015. This identification method, known as the staggered adoption design, has recently been the subject of numerous methodological developments.
In particular, it relies on the assumption of a constant treatment effect, both over time and across cohorts. To relax these assumptions, the study compares the results of the standard estimation method (two-way fixed effects) with three alternative estimators developed by De Chaisemartin and d’Haultfoeuille (2020), Callaway and Sant’Anna (2021) and Gardner (2021). The estimated effects are found to be very similar, regardless of the method used.
Data used
The study draws on annual censuses of pupils enrolled in secondary education (FAERE data from the SCOLARITÉ information system) and students enrolled in higher education (STS/CPGE data from SCOLARITÉ and SISE data) over the period 2006–2018.
These data were provided by the Directorate for Evaluation, Forecasting and Performance of the Ministry of National Education and Youth (MENJ-DEPP) and the Sub-Directorate for Information Systems and Statistical Studies of the Ministry of Higher Education and Research (MESR-SIES).
It provides detailed information on the socio-demographic characteristics of pupils and students (age, gender, nationality, municipality of residence, socio-professional category of legal guardians), as well as the institution attended and the course of study undertaken each year. These data are supplemented by A-level results, drawn from the OCEAN information system, as well as pre-enrolment choices for higher education, drawn from the Admission Post-Bac (APB) platform for the year 2015.
The national pupil identification number (INE), in encrypted form, enables these different sources to be linked in order to reconstruct pupils’ educational pathways, from their final year of secondary school until they leave the education system. Finally, geolocation data for secondary and higher education institutions, as well as information on the size and type of local authorities, are sourced from open data.