真实固定效应空间随机前沿模型的贝叶斯估计

忽略个体效应和空间效应会严重干扰效率测算，其中忽略个体效应使得技术无效率项发生偏移，忽略空间相关性导致估计量有偏且不一致。本文基于真实固定效应随机前沿模型（引入了个体效应），引入因变量和双边误差项的空间滞后项，构建了适用性更佳的真实固定效应空间随机前沿模型。对模型进行组内变化以消除额外参数，使用贝叶斯方法（需推导未知参数的后验分布并执行MCMC抽样）估计参数和技术效率。该方法真正克服了额外参数问题，比同类方法直观、简便。数值模拟结果表明，本文方法对参数、个体截距项及技术无效率项的估计精度均较高,且增加样本容量，估计精度变优。

The Bayesian Estimation of True Fixed Effects Spatial Stochastic Frontier Models

Ignoring individual effects and spatial effects would seriously affect the estimation of technical efficiencies, ignoring individual effects will shift the technical inefficiency items and ignoring spatial effects will result on the bias and inconsistent estimators. This paper simultaneously incorporates the spatial lagged terms of dependent variable and stochastic errors into true fixed effects stochastic frontier models, then construct the true fixed effects spatial stochastic frontier models which are more applicable. The changes within the model to remove additional parameters and use Bayesian method (need to deduce posterior distribution and execute MCMC sampling) to estimate parameters and technical efficiencies. The method truly overcomes the additional parameter problem which is simpler and more intuitive. The numerical simulations show that the estimation accuracies of parameters, individual effects and technical efficiencies are high, and with the sample size increasing, the estimation accuracies become higher.