混合性(Mixing)在时间序列和空间计量经济学研究中起着重要的作用,许多时间序列文献都假设其模型中的变量服从混合过程.然而,目前尚无关于判定空间计量经济模型所生成数据是否满足混合性质的准则.基于Doukhan(1994)的思想,基于若干常见假设,我们建立了一系列准则,用于判定不规则格点上的线性空间过程是否满足 α-混合性.我们将这些准则应用于建立由空间自回归模型、空间误差模型、矩阵指数空间模型以及基于潜在被解释变量的空间计量经济模型(例如空间样本选择模型)所生成被解释变量的α-混合性质.利用α-混合性质,我们建立了Flores-Lagunes et al.(2012)提出的空间样本选择模型的估计量的大样本性质.
Conditions for α-Mixing Linear Random Fields:With Applications to Spatial Econometric Models
Mixing plays a fundamental role in time series and spatial econometrics,and many time series papers assume that the variables in their models follow a mixing process.However,in the literature on spatial econometrics,there are no criteria to establish the mixing property for spatial econometric models.Following the general idea in Doukhan(1994),based on some common assumptions,we establish some criteria for a linear spatial process over an irregular lattice to be α-mixing.We apply the criteria to establish the α-mixing property generated by the spatial autoregressive model,the spatial error model,the matrix exponential spatial model,and spatial econometric models with qualitative and limited dependent variables based on latent dependent variables,such as a spatial sample selection model.Using the α-mixing property,we establish large sample properties of estimators for the spatial sample selection model proposed in Flores-Lagunes et al.(2012).
α-mixingspatial autoregressive modelspatial error modelspatial sam-ple selection model
NETL
NSTL
万方数据
