Experimental Design
We will send out fictitious CVs and cover letters to vocational training programs and employers. These applications relate to six low-skilled trades: care assistant, secretary, forklift operator, employees in restaurants, sales clerk and truck driver. We will use six age categories (35, 40, 45, 50, 55, and 60). For each occupation, we will create two types of profiles. One will represent a rather highly educated profile (at least high school) and one a rather low educated profile (below high school diploma). By observing average differences in the rate of positive responses according to the variation introduced, we can identify the causal effect of this variable on the response rate.
In the first part (applications to training providers), we send two applications to the same training program. The two applicants differ by age. Further, we make sure that the two applications also differ in other dimensions which are not related to the age of the applicants (e.g. name, template of the CV, content of the email message), to avoid potential detection by the training provider. In the second part (applications to potential employers), we send one or two applications to each potential employer. The applications differ by age and by the fact whether the applicant has completed a training. Again, we make sure that the applications also differ in other dimensions which are not related to the age of the applicants (same dimensions as in the applications to training providers), to avoid potential detection by the employer.
Our design allows us to measure (i) the effect of the age of the applicant on the probability of a positive response from the training center and (ii) the age-specific effect of vocational training on the probability of a candidate receiving a positive response to his or her application.
In addition, we conduct an online survey of training providers. The survey has been launched on June 12, 2026. The survey collects information on the characteristics of training participants, including the age distribution of trainees, and on the procedures used to select candidates for training programs. In particular, respondents are asked to assess the importance of several candidate characteristics in admission decisions, including motivation, the quality of the professional project, formal educational attainment, prior work experience, learning abilities, the capacity to keep pace with the training program, digital skills, and expected employment prospects after training.
The survey also elicits training providers’ perceptions of workers from different age groups. Respondents evaluate age groups with respect to speed of learning, ease of pedagogical support, ease of integration into trainee groups, likelihood of dropping out before completing training, and expected ease of labor market insertion after training completion.
Finally, the survey gathers information on the incentives faced by training providers. Specifically, respondents are asked whether the funding of their organization depends on trainee attendance, whether funding bodies place importance on post-training employment outcomes, and whether public funding contracts (e.g., public procurement contracts, France Travail agreements, or regional funding arrangements) include quantitative targets related to participants’ employment placement rates.
Combined with the experimental evidence, these survey data will be used to investigate the mechanisms underlying age-related differences in access to training. In particular, they will allow us to distinguish between two broad explanations for the documented age penalty: (i) institutional incentives that encourage training providers to prioritize applicants with stronger expected employment prospects and (ii) providers’ beliefs about age-related differences in learning capacity, training completion, and the impact of training on subsequent employability.
Whenever possible, survey responses will be linked to the training providers included in the correspondence study. This linkage will allow us to relate providers’ reported selection criteria, perceptions of applicants of different ages, and institutional incentives to their observed responses to fictitious applications. By combining survey and experimental data at the provider level, we will investigate whether providers exhibiting larger age penalties in the correspondence study are also more likely to report stronger concerns about the employability of older workers or stronger incentives tied to employment placement outcomes.