关键词: 部分线性变系数面板模型, 固定效应, 最小二乘支持向量机, 二次推断函数法

Abstract: This paper provides fast efficient estimators for a partially linear varying coefficient panel model with fixed effects and relevant structures among units by conflating the least square support vector machine (LSSVM) and the quadratic inference function method (QIF). Then an asymptotic normality of the parameter estimators and a convergence rate of the non-parameter estimators are derived in some regular conditions. At the end, a finite sample performance of the proposed method is observed via Monte Carlo simulations, and a real data set is applied to this proposed estimation technique. It is found that the efficiency of estimators and statistical inference ability are ensured and moreover the running speed of the program is greatly enhanced.

Key words: Partially Linear Varying Coefficient Panel Model, Fixed Effects, Least Square Support Vector Machine, Quadratic Inference Function Method