Efficient L 1 estimation and related inferences in linear regression with unknown form of heteroscedasticity

摘要: In linear regression analysis, L 1 estimation of parameters is a viable alternative to the least squares estimation. It an extension concept median, thereby having certain desirable robustness properties, and can be easily implemented via programming (Koenker Bassett, 1978). Like method, it lose efficiency when error terms are heteroscedastic. Koenker Zhao (1994) showed that, parametric form heteroscedasticity assumed, one obtain asymptotically efficient estimator by reweighting equation, where optimal weights consistently estimated. this article, we consider unknown. We propose sample-splitting method construct consistent estimates for avoid bias. make use recently developed resampling approximate sampling distribution resulting estimate. Simulation results which provide assessment ef...