Título: “Essays on microeconometrics and impact evaluation in the context of education and labor outcomes”

Autor: Marco Aurelio Pérez Navarro
Directoras: Rocío Sánchez-Mangas y Maite Blázquez Cuesta
Universidad: Universidad Autónoma de Madrid

Fecha de defensa: 23/03/2026
Comité: Manuel Bagüés, Pilar Poncela y César Alonso-Borrego. 

Calificación: Sobresaliente Cum Laude

This dissertation combines methodological advances in microeconometrics and the use of impact evaluation techniques, to analyze several aspects in the context of education and labor economics. The structure of the thesis consists of an introductory chapter, three core chapters written in a self-contained but interconnected manner that maintain thematic coherence, and a final chapter that offers the concluding remarks.

The second chapter examines misspecification issues in trivariate probit models with recursive structures and sample selection. More specifically, we analyze the consequences that the omission of recursive dependencies has on the correlation coefficients of the error terms of the model equations. We show analytically that this omission can lead to important biases in the estimated correlation coefficients and to wrong conclusions about the existence of sample selection. We contribute to the literature by extending the work by Filippini et al. (2018). These authors focus misspecification in recursive bivariate probit models, while our work incorporates sample selection as an additional source of endogeneity, leading to a trivariate probit model in which a double recursive structure is considered. To illustrate our findings, we present the results form a Monte Carlo simulation exercise.

The third chapter analyzes the phenomenon of overeducation among recent university graduates in Spain. Using data from the 2014 and 2019 waves of the Survey on the Labor Insertion of University Graduates (EILU), we investigate the role played by the business cycle and the field of study and their interaction in shaping both the incidence and persistence of overeducation. The modeling strategy, based on the findings in the previous chapter, considers several endogeneity sources: endogenous regressors, individual unobserved heterogeneity and sample selection into employment. The results show that graduates entering the labor market during periods of recession face a higher risk of overeducation and of its persistence over time, with significant heterogeneity across fields of study. Furthermore, a complementary analysis highlights the compensatory role of transversal skills in reducing this risk.

The fourth chapter presents the evaluation of an Active Labor Market Policy (ALMP) implemented in Madrid in 2019, targeting long-term unemployed individuals with prior formal education. The program combines six months of subsidized employment in the public sector, with training in cross-cutting skills with a strong focus on employability and job-search assistance. Using administrative microdata complemented with additional sources, we estimate the causal effect of the program on the probability of obtaining a job that matches the individual’s educational level. We contribute to the scarce existing evaluations of subsidized public employment programs in Spain and we focus on a labor outcome not studied in the ALMP literature. The results indicate an overall positive effect of small magnitude, which decreases over time. However, a clear heterogeneity pattern across educational levels emerges: Higher Vocational Education and Training (VET) graduates benefited significantly across all the studied period, whereas the effects for university graduates and graduates of Intermediate VET were statistically insignificant, suggesting the need for a policy redesign for these groups. A sensitivity analysis is performed to support the robustness of the results to the existence of potential unobserved confounders.

Overall, this dissertation examines various aspects of the interrelation between the economics of education and labor, combining empirical analysis with methodological advances in microeconometrics. It highlights the importance of properly modeling complex recursive structures, while also providing new evidence on overeducation, the role of skills, and the effectiveness of labor market interventions.