摘要
经典的随机前沿模型忽略了决策单元之间的空间关联性,无法准确估计效率影响因素相关参数,限制了其适用范围。本文在空间自回归随机前沿模型的基础上,引入效率影响因素,构建出一个异质性空间随机前沿模型,基于极大似然估计法给出模型参数的单步估计策略,提出决策单元技术效率的最优预测量。理论分析证明,模型参数在一定的假设条件下具备一致性;模拟实验表明,参数估计量和技术效率预测量较之经典模型具有更高的估计精度,且会随着样本量的扩大而逐渐提升。本文使用所提出理论方法讨论了我国城市数字普惠金融发展与技术效率水平之间的相关关系,发现两者之间存在显著的正相关关系,同时也印证了模型设定和估计方法的可靠性。
Classic stochastic frontier models ignore the spatial relationship between decision-making units(DMUs) and cannot accurately estimate corresponding parameters of the efficiency influencing factors,which limits their scope of application.This paper brings those factors into the spatial autoregressive stochastic frontier model,constructs a heterogeneous spatial stochastic frontier model,proposes a single-step estimation strategy for model parameters and an optimal technical efficiency predictor of DMUs based on maximum likelihood estimation.Theoretical analysis proves that the model parameters are consistent under certain assumptions;simulation experiments show that the parameter estimators and technical efficiency predictor have higher accuracy than classic models,and the accuracy continues to improve as the sample size increases.The case application uses the theoretical method proposed in this article to discuss the correlation between digital financial inclusion and the technical efficiency in China’s cities,finds that there is a significant positive correlation between the two and verifies the reliability of the model specification and the estimation method.
作者
冯冬发
张涛
李奥
宫汝娜
Feng Dongfa;Zhang Tao;Li Ao;Gong Runa
出处
《统计研究》
CSSCI
北大核心
2023年第1期144-156,共13页
Statistical Research
基金
国家自然科学基金重大项目“宏观大数据建模和预测研究”(71991475)
中国社会科学院大学研究生科研创新支持计划项目“大数据在宏观经济预测和政策评价中的应用”(2020-KY-048)。
关键词
空间自回归
随机前沿模型
技术效率
极大似然估计
Spatial Autoregression
Stochastic Frontier Model
Technical Efficiency
Maximum Likelihood Estimation
