空间滞后分位数回归模型的贝叶斯估计

Bayesian Estimation of Spatial Lag Quantile Regression Model

提出空间滞后分位数回归模型的贝叶斯估计方法。研究方法:根据贝叶斯分析思想,分别在模型参数的正态先验和双指数先验设定下,构建模型参数的贝叶斯估计方法,并分别利用数值模拟方法和应用实例考察估计方法的小样本表现和实际应用效果。研究发现:所提出的贝叶斯估计方法在小样本条件下具有良好的估计效果和稳健性,在两种先验设置下,不同分位点上的参数估计精度均较高,应用实例展示了理论方法的实际应用价值。研究创新:应用贝叶斯方法估计空间分位数回归模型,该方法综合考虑了先验信息和样本信息,具有更高的估计精度。研究价值:所构建的理论方法将为经济、金融、环境等领域的具有厚尾和空间相依特征的数据提供有力的分析工具。

Research Objectives:To propose bayesian estimation method for spatial lag quantile regression modelResearch Methods:Construct bayesian estimation method under normal prior and double-exponent prior setting of model parameters respectively according to bayesian analysis idea Use the numerical simulation method to examine the small sample performanceUse practical example to investigate the practicality of methodResearch Findings:The proposed estimation method has good accuracy and robustness under small sample conditionsUnder the two prior settings of model parameters, the estimation accuracy at different quantile is highThe practical example shows the application value of the theoretical method Research Innovations:To apply bayesian estimation method to analyze spatial quantile regression modelThe method considers a prior information and sample information comprehensivelySo the estimation accuracy is improvedResearch Value:This paper provides an effective tool for the data with fat tail and cross-section dependence in various research areas, such as Economics, Finance, Environmental Science, and so on.