NA误差下半参数回归模型估计的相合性

摘要

For semiparametrie regression model Yβ(xin, tin)=tinβ+g(xin)+eβ(xin), when the errors eβ(xin) are negatively associated random errors, we study the strong consistency and rth mean (r ＞ 2) consistency for the estimators βmin and gmin(x) of β and g, respectively. The results of Hu[2] on independent random errors are extended and the restrictions for weight function Wni(x) are weakened.%对半参数回归模型: Y(j)(xin,tin)=tinβ+g(xin)+e(j)(xin),1≤j≤m,1≤i≤n,本文在NA相依样本下讨论了g的加权估计及β的最小二乘估计的强相合性与r(＞2)阶平均相合性,使得文献[2]在独立样本下的相应结果得到推广.