線性迴歸模型及聯立方程式模型之Mallows形式有界迴歸分位向量

Abstract

我們導出線性迴歸模型及聯立方程式模型的Mallows形式有界迴歸 分位向 量之近似分配。我們也做蒙地卡羅(Monte Carlo)模擬，並比 較均方誤差且 驗證出如果獨立變數產生毛差(gross errors)時，則有界 迴歸分位向量將比無界迴歸分位 向量(見Koenker及Bassett(1978))來得 有效率 。 We present asymptotic distributions of the Mallows type bounded-influencereg ression quantile for the linear regression model and also the simultaneousequa tions model. Monte Carlo simulation comparing mean squared errors showsthat th e bounded-influence one is more efficient than the unbounded- influence one(Koe nker and Bassett(1978))when gross errors occur in the independent-variables-s pace.