Robust penalized quantile regression estimation for panel data

摘要: This paper investigates a class of penalized quantile regression estimators for panel data. The penalty serves to shrink vector individual specific effects toward common value. degree this shrinkage is controlled by tuning parameter λ. It shown that the asymptotically unbiased and Gaussian, when are drawn from zero-median distribution functions. parameter, λ, can thus be selected minimize estimated asymptotic variance. Monte Carlo evidence reveals estimator significantly reduce variability fixed-effect version without introducing bias.