Semi and Non parametric Econometrics

Download the full article

We review the theoretical foundations and the most important properties of the quantile fixed-effect estimator suggested by Koenker (2004) to handle panel data. This estimator was specifically designed to take into account the panel structure of datasets when the econometrician wants to measure heterogenous effects of the covariates using quantile regression. We apply it on US wage data, and finds that it achieves a good performance in terms of variance reduction without introducing too much bias. However, we underline two weaknesses, related to its asymptotic properties and to the tuning of the penaly parameter.

This work was realized in the context of the Semi and Non-Parametric Economics cours at ENSAE.