基于季节增长率的一种直接调整方法

发布时间：2019-05-13 07:53

【摘要】：传统季节调整方法在提取环比增长率时需要先剔除原始数据中的季节成分,这会带来原始数据信息的失真。鉴于此,本文提出了一种直接拟合原始数据增长率的季节增长率(SGR)模型,该模型不仅可以直接提取环比增长率,还可以对原始数据的增长率进行预测。蒙特卡洛模拟结果表明,本文给出的针对SGR模型的MLE估计方法具有良好的有限样本表现。通过对我国GDP和CPI数据进行实证,本文发现利用SGR模型直接提取的环比增长率的稳定性要高于其他一些季节调整方法。不仅如此,SGR模型的拟合和预测表现相比BSM模型和SARIMA模型均有显著提高。此外,SGR模型还具有容易拓展为非线性、多元情形的优势。
[Abstract]:The traditional seasonal adjustment method needs to remove the seasonal components from the original data when extracting the annular growth rate, which will lead to the distortion of the original data information. In view of this, this paper proposes a seasonal growth rate (SGR) model which can directly fit the growth rate of raw data. The model can not only directly extract the annular growth rate, but also predict the growth rate of raw data. Monte Carlo simulation results show that the MLE estimation method for SGR model presented in this paper has good finite sample performance. Through the empirical study of GDP and CPI data in China, this paper finds that the stability of ring specific growth rate extracted directly by SGR model is higher than that of other seasonal adjustment methods. Moreover, the fitting and prediction performance of SGR model is significantly better than that of BSM model and SARIMA model. In addition, the SGR model also has the advantage of easily expanding to nonlinear and multivariate cases.
【作者单位】：厦门大学经济学院统计系;厦门大学王亚南经济研究院;
【基金】：长江学者奖励计划青年学者项目(Q2016131) 国家自然科学基金面上项目“状态空间混频模型及其在宏观经济中的应用(71371160)” 教育部新世纪优秀人才支持计划“混频数据建摸及其在经济中的应用(NCET-13-0509)” 厦门大学基础创新科研基金“利率期限结构的混频建模及应用(201422G010)”的资助
【分类号】：F224

【参考文献】