This paper shows that in finite samples it is not possible to distinguish all the latent factors from the idiosyncratic noise and that this leads to a bias towards the identification of a single factor. It provides an approximation to this bias and the corresponding sampling distribution.

This paper proposes an ℓ1 penalized quantile regression estimator which adapts the Hausman–Taylor instrumental variable approach in order to address the bias resulting from the shrinkage of the individual effects.

We introduce a quantile regression approach to panel data models with endogenous variables and individual effects correlated with the independent variables. We find newly developed quantile regression methods can be easily adapted to estimate this class of models efficiently.

We propose a method for estimating the slope parameter in an interactive effects panel data model with endogenous loadings and factors, and endogenous regressors.