固定效应部分线性可加面板数据模型的惩罚二次推断估计

具有较强解释力和灵活性的部分线性可加面板数据模型在各学科领域应用广泛.针对个体内存在相关结构的固定效应部分线性可加面板数据模型,本文在结合幂样条函数和最小二乘虚拟变量(LSDV)法的基础上,利用惩罚二次推断函数(PQIF)法对模型进行估计,在一定的正则条件下,证明了参数估计的渐近正态性和非参数估计的收敛性,Monte Carlo数值模拟显示所述估计方法具有良好的有限样本表现,同时,我们还将估计技术应用于实际数据分析中.

Penalized Quadratic Inference Estimation of Fixed Effects Partially Linear Additive Panel Data Model

Partially linear additive panel data models with strong interpretability and flexibility have been widely used in a variety of research fields. Considered a fixed effects partially linear additive panel data model with correlation structure within subjects, we derived the estimators by using penalized quadratic inference functions method under the basis of combining exponential spline function and LSDV method; the asymptotic normality of parametric estimators and convergence of nonparametric estimators were proved under suitable regular conditions; Monte Carlo simulations show that our estimates have good performances in small sample cases; meanwhile, the estimation techniques were used to analyse a real data set.