连续性抽样的时间序列估计方法研究进展及应用

自1973年Blight和Scot首次提出关于连续性抽样调查的时间序列估计方法以来,国际上有关时间序列估计方法的研究发展得十分迅速。本文选择时间序列估计方法作为研究对象,主要从利用经典时间序列理论进行抽样估计、利用卡尔曼滤波进行抽样估计、时间序列估计方法的实际应用三方面对现有研究成果进行了理论化、系统化的梳理。同时,结合现有研究中存在的模型选择、假定条件合理性等问题以及未来研究趋势,建议我国统计调查部门应加快统计调查活动科学化、规范化的进程,从市县层级调查活动入手,在实践中不断完善时间序列估计理论。

The theoretical research on time series estimation method of repeated sampling survey has been developed rapidly in the world since Blight and Scott firstly raised this estimation method in 1973. Taking time series estimation method as the research object, this paper makes a theoretical and systematic review of the existing research results in three aspects： using classic time series estimation for sampling estimation, using Kalman filter for sampling estimation and making practical application of time series estimation method. In addition, combined with the existing problems of model selection and the rationality of hypothetical conditions and the research trend in the future, the author suggests that the statistical investigation departments in China should speed up the process of scientific and standardized statistical investigation activities, start with the investigation activities at the city and county levels, and improve the theory of time series estimation in practice.