广义空间模型的方差齐性检验

TESTING FOR HETEROSCEDASTICITY IN GENERAL SPATIAL MODELS

研究了广义空间模型的方差齐性检验问题,在异方差情形下导出了Score检验统计量的具体形式和近似分布.分别应用于混合空间自回归模型和空间误差模型,给出了相应的检验统计量和渐近分布.并利用Monte Carlo模拟对检验统计量的性质进行分析.最后,通过中国能源利用效率的区域特征数据证明了方法的有效性.

This paper studies the score test for heteroscedasticity in general spatial model,which including spatial autoregressive models（SAR） and spatial error models（SEM） as special cases.Score test statistic and its approximate distribution are derived under the heteroscedastic model.Then the properties of test statistic are investigated through Monte Carlo simulations.Finally,we illustrate our method using a real data of the regional characters of energy efficiency in China.