解释变量内生假定下非参数空间计量模型的局部线性工具变量估计

Local Linear Instrumental Variable Estimation of a Nonparametric Spatial Model Under Endogenous Explanatory Variable Assumption

为刻画解释变量的空间外溢效应与变量间的非线性关系,本文提出了一种非参数空间计量模型,并给出了模型含内生变量情况下的局部线性工具变量估计。该估计方法最大的优点是可以同时获得偏导数的估计,便于进行经济学的边际分析。数值模拟结果表明,局部线性工具变量估计优于核估计。中国地区R&D要素外溢效应的实证结果显示,周边地区R&D内部经费支出对本地区R&D产出具有非线性的正向影响,且周边地区R&D内部经费支出的边际产出存在空间集聚现象,实证结论显示了非参数空间计量模型的适用性与合理性。

This paper proposes a nonparametric spatial model to capture the spatial spillover effect of explanatory variable and the nonlinear relationship in variables, as well as proposes a local linear instrumental variable estimation for the nonparametric spatial model with endogenous explanatory variable. As an advantage of this estimation method, it can estimate the partial derivative, which facilitates marginal analysis in economics. The numerical simulation results show that local linear instrumental variable estimation is superior to kernel estimation. The empirical results of R＆D element spillover effect show that R＆D intramural expenditure of neighborhood regions has a nonlinear positive influence on R＆D output, and the marginal output of R＆D intramural expenditure from neighborhood regions appears spatial agglomeration phenomenon. The empirical results also validate the applicability and rationality of the nonparametric spatial model.