Skip to Main content Skip to Navigation

Essai sur quelques problèmes d'identification en économie

Abstract : This PhD thesis presents three independent research topics, related however by the issue of the identification of economic models. The first chapter is dedicated to the identification of nonparametric instrumental models. I first study the completeness condition, which has been recently used in nonparametric instrumental regression or in measurement error models. For that purpose, I suppose that a nonparametric, additively separable model holds between the two variable, together with a large support condition. In this framework, different versions of completeness are obtained, depending on the regularity conditions imposed on the model. The second part of this chapter develops a new method for dealing with endogenous selection. This method is based on the independence between the instruments and the selection and relies on a completeness condition between the outcome and the instrument. An estimation procedure and an application are also proposed. The second chapter studies two empirical industrial organization models. The first part considers the nonparametric identification of the common value auction model. The main identifying assumption is that the support of the distribution of the signals, conditional on the value of the good, varies with this value. The advantage of our approach is that, apart from this condition, it does not rely on functional restrictions, contrarily to the existing literature. The second part focuses on the adverse selection model. This model is defined by the objective function of the principal, the agent's utility and the distribution of their type. We show that the identification of the model requires the knowledge of at least one of the three functions. We also show that exogenous changes in the objective function of the principal enable to identify fully or partially the model. A nonparametric estimation is proposed and used to test the optimality of contracts. The third chapter focuses on peer effect models. Whereas these models are usually considered nonidentified, we show that a slight modification of the standard linearin-means model enables in general to identify the structural parameters, by using the group size variations. These results are extended to a binary version of the model. Parametric estimation of the model is also considered, as well as the finite distance properties of the estimator.
Document type :
Complete list of metadata

Cited literature [176 references]  Display  Hide  Download
Contributor : Lucie Label Connect in order to contact the contributor
Submitted on : Wednesday, July 8, 2009 - 4:43:04 PM
Last modification on : Friday, August 5, 2022 - 2:49:41 PM
Long-term archiving on: : Tuesday, June 15, 2010 - 7:45:55 PM


  • HAL Id : tel-00402960, version 1



Xavier d'Haultfoeuille. Essai sur quelques problèmes d'identification en économie. Economics and Finance. Université Panthéon-Sorbonne - Paris I, 2009. English. ⟨tel-00402960⟩



Record views


Files downloads