统计研究

• 论文 • 上一篇    下一篇

大数据指数是否可以替代统计调查指数

米子川 姜天英   

  • 出版日期:2016-11-15 发布日期:2016-11-11

Whether Big Data Index Can Replace Survey Data Index

Mi Zichuan & Jiang Tianying   

  • Online:2016-11-15 Published:2016-11-11

摘要: 2014年7月,澳盛银行首次将阿里巴巴系列指数纳入通胀观察标的,标志着大数据指数已经开始对传统的统计调查指数提出质疑和挑战。本文基于阿里巴巴aSPI指数和官方公布的CPI指数的比较研究,首次提出了aSPI指数显著优于CPI指数的一些基本特征;同时,通过实证分析对比了两种指数的同步性特征和分解性特征,即首先运用协整检验方法确定二者的同步性;其次通过EMD模型对二者进行序列分解,得出各自的波动成分和增长趋势;最后,在EMD对aSPI指数分解的基础上,通过Lasso回归估计了CPI指数。研究表明,随着对大数据研究的广泛性、科学性以及方法论和软件工具的进步,大数据指数对传统统计调查的佐证、补充乃至融合将会成为一种新趋势,通过实证、应用与发展,逐步产生新的CPI编制方法和分析体系,将是大数据指数理论和实践的根本出路。

关键词: 大数据指数, 抽样调查指数, EMD-Lasso模型

Abstract: Based on the comparative study of aSPI and CPI, this paper firstly proposed some superior basic characteristics of aSPI in comparison with CPI. Simultaneously, by carrying with empirical analysis, features in terms of synchronization and decomposition can be contrasted between these two indices, which is using cointegration model to ensure synchronization. Secondly, we take use of series decomposition with EMD to get the random and trend components separately. At last, based on the previous results on aSPI, capitalizing Lasso regression to estimate the CPI. The research shows that, with the development of big data research, scientific data and the progress of software tools, the support of big data index to traditional statistical survey will be a new trend. It will be the fundamental way out of the big data index theory and practice which gradually produces a new CPI preparation methods and analysis system through examination, application and development.

Key words: Big Data Index, Sampling Survey Index, EMD—Lasso Model