Doctoral thesis defense of Álvaro de Jesús Fernández Junquera

This doctoral dissertation aims to provide evidence for the design of active labour market policies (ALMPs). The research is structured into two main objectives: evaluating the effectiveness of specific design features of ALMPs on employment outcomes and assessing jobseeker profiling models to optimize the selection of target groups.

To address the first objective, we developed two impact evaluations. In Chapter II, we investigated the role of timing, intensity, and occupational specificity in an adult training program in Catalonia (Spain). We employed a conditional difference-in-differences strategy and found that the overall employment effect of the program was positive. However, we did not find clear evidence of a timing effect, as we obtained minor differences across cohorts. We detected a remarkable non-linear effect of the intensity of work-based learning (internships) on the employment probability. Nonetheless, the impacts of the intensity of classroom-based learning and occupational specificity were substantially smaller. 

In Chapter IV, we analysed the role of job search assistance and intermediation in a program implemented through vouchers and performance-based pay in Veneto (Italy). The policy assigned individuals to three treatment groups with different treatment endowments according to the score of a profiling model. Using a regression discontinuity strategy, we compared the effects of different treatment levels and did not find an impact on the employment outcomes after two years. Regarding the assistance component, one key finding was that the vouchers were underutilized, leading to minimal differences in treatment uptake. Concerning the intermediation component, the differences between the payments of each treatment group seemed overall insufficient to yield lasting effects.

To achieve the second objective, we studied the predictive performance of two types of jobseeker profiling models. The rule-based model currently deployed in the public employment service of Catalonia was found to have poor predictive performance, both in its ability to discriminate jobseekers that will be long-term unemployed and in calibrating the employment predictions. In contrast, a newly developed statistical model demonstrated superior performance, leveraging administrative data and machine learning techniques to achieve excellent calibration. This model could potentially better align ALMP resources with individual needs.

Our research has tried to contribute both to the academic and the policy spheres. With respect to the academic literature, we have paid greater attention to the specific tools and target groups included in program designs. This approach allows us to go beyond the simple classification of types of ALMPs, digging into the components of the intervention that might attain a behavioural or cognitive change. Moreover, we have advanced contemporary research on different topics like the occupational specificity or the voucher implementation by studying them in the framework of ALMPs. With respect to policymaking, we have brought evidence on highly discussed issues in Southern Europe like the profiling of jobseekers or the use of internships in adult training programs. Both elements have been recently regulated at the national level in Spain and Italy, so our analyses might serve to continue the policy design at the rest of government levels.

 

Date: 25/10/2024   11,30h

Sala de Graus (Fac. Ciències Econòmiques)