2 juillet, Soutenance de thèse de Jasmin FLIEGNER

2 Juillet 2019 Recherche

Madame Jasmin FLIEGNER soutiendra sa thèse de doctorat en Sciences économiques le 02/07/2019 à 15h00, Salle  MS 001   (Manufacture des Tabacs)

Sur le sujet : « Essays on the Econometrics of Program Evaluation: On the Reliability of Observational Methods based on Uncounfoundedness»
Directeur de thèse: Pascal LAVERGNE, Professeur TSE
Le jury se compose comme suit :

  • Professeure Anne VANHEMS, TSE
  • Professeur Christoph ROTHE, Université de Mannheim
  • Professeur Marc GURGAND, Directeur de Recherches CNRS, PSE
  • Professeur Axel WERWATZ, T.U Berlin
  • Professeur Sylvain CHABE-FERRET, TSE
  • Professeur Pascal LAVERGNE,TSE


Résumé: (en anglais)

This dissertation makes a contribution to the econometrics of program evaluation and their applications. When it comes to the question to analyze the causal effect of a program or a binary treatment on a set of outcomes, several statistical tools have been developed to overcome the problem of selection bias.
Once the research question can be put in the form of a thought experiment as suggested by Richard Feynman2, actual experiments seem the most straight forward way and a strong statistical tool to identify a causal effect. However, the plurality of queries in economics cannot always be addressed by such an experiment due to budget or moral- ity concerns or infeasibility. Therefore, it is advantageous to have reliable observational methods as well to tackle causal questions. My current research aims at understanding and improving the reliability of observational methods. In this dissertation I consider  the meaning of the word reliability to be twofold. Two chapters of this dissertation focus on the assessment of the quality of observational estimation methods based on the un- confoundedness assumption. Therefore the term reliable refers to the methods’ ability to overcome the selection bias problem to identify the causal effect given that they mostly rely on stronger assumptions compared to experimental methods. However, having iden- tified the causal effect is of little value without the possibility to draw conclusions about the statistical significance of the resulting estimate. The last chapter of my thesis pro- vides an inference procedure for estimators based on the unconfoundedness assumption.