空间滞后分位数回归模型的截面估计法

Profile Estimation of Spatial Lag Quantile Regression Model

研究目标:提出空间滞后分位数回归模型并构建其估计方法。研究方法:根据截面估计思想构建模型的估计方法,分别利用数理推导和蒙特卡洛模拟方法得到估计量的大样本性质和估计方法的小样本表现,并通过应用实例考察理论方法的实用性。研究发现:利用所构建估计方法得到的估计量具有一致性和渐近正态性,且估计方法在小样本条件下具有良好的估计精度和稳定性,实际应用结果也展示了模型的应用价值。研究创新:本文提出的模型不仅考虑了因变量的截面相依性,还能考察自变量对因变量整个条件分布的影响,所构建的估计方法可以避免传统方法中工具变量的选择问题,估计精度更高。研究价值:本文构建的理论方法将为分析经济、金融、环境等领域中常见的具有厚尾特征的数据提供强有力的研究工具。

To propose spatial lagged quantile regression model and construct its estimation method.Research Methods:Construct the model estimation method based on profile method,use mathematical derivation and Monte Carlo simulations methods to obtain the large sample properties of the estimators and the small sample performance of the estimation methods respectively.Research Findings:The Estimators have consistency and asymptotic normality and good accuracy and stability in small sample conditions.The practical application results also show the application value of the model.Research Innovations:The model not only considers the cross-section dependence of the dependent variable,but can also examine the effect of independent variables on the whole conditional distribution of dependent variable.At the same time,the proposed estimation method can avoid the problem of instrument variables selection in traditional methods and improve estimation accuracy.Research Value:This paper provides an effective tool for the data with fat tail in various research areas,such as Economics,Finance,Environmental Science,and so on.