资源科学 ›› 2016, Vol. 38 ›› Issue (10): 1821-1831.doi: 10.18402/resci.2016.10.01

• 海洋资源 •    下一篇

基于混频MF-VAR模型的中国海洋经济增长研究

殷克东1, 金雪1, 2, 李雪梅1, 王凤娇1   

  1. 1. 中国海洋大学经济学院,青岛 266100;
    2. 中国海洋大学海洋与大气学院,青岛 266100
  • 收稿日期:2016-02-22 修回日期:2016-08-29 出版日期:2016-10-25 发布日期:2016-10-25
  • 通讯作者: 金雪,E-mail:qzhjinxue2008@163.com
  • 作者简介:殷克东,男,山东临沂人,教授,博士生导师,主要研究方向为数量经济分析与建模、海洋经济计量分析。E-mail:yinkedong@126.com
  • 基金资助:
    国家社会科学基金重大项目(14ZDB151); 国家海洋公益项目(201405029-1); 教育部哲学社会科学发展报告培育项目(13JBGP005)

Growth in China's marine economy based on MF-VAR model

YIN Kedong1, JIN Xue1, 2, LI Xuemei1, WANG Fengjiao1   

  1. 1. School of Economics,Ocean University of China,Qingdao 266100,China;
    2. College of Physical and Environmental Oceanography, Ocean University of China,Qingdao 266100,China
  • Received:2016-02-22 Revised:2016-08-29 Online:2016-10-25 Published:2016-10-25

摘要: 由于中国海洋经济相关指标月度数据和季度数据未定期公布,所以以年度低频海洋经济样本数据进行分析时会因样本数据过短而结果不精确。本文构建了海洋经济增长混频MF-VAR模型,根据2005-2013年中国海洋经济增长实时数据对模型进行最优选取和参数估计,与基准模型进行测度对比,探索混频数据模型在海洋经济领域的应用。研究结果表明:①MF-VAR模型在测度中国海洋经济增长方面误差相对较小,且多变量MF-VAR模型拟合效果优于GOPCIFAGOPVFH的单变量MF-VAR模型,说明海洋经济的周期波动受到各方面因素的影响,仅是影响程度大小不同;②相对于基准模型,MF-VAR模型在短期预测方面具有精准性的比较优势,随着预测步数的增加,估计和预测精度下降;③不论是单变量还是多变量MF-VAR模型,估计和预测的MSE均低于对应同频数据基准模型,因此混频数据模型不仅可以解决样本长度较短的问题,而且在提取海洋经济高频数据信息方面具有显著优势,可提高海洋经济分析的准确度和及时性。

关键词: MF-VAR模型, 海洋经济, 混频数据, 基准模型

Abstract: Since the 21st century,the development of China's marine industry has ushered in a new era and the marine economy has become an important part of China's macro-economy. Formulation of national macroeconomic policy and the implementation of marine economic strategy all need to have a comprehensive grasp of the current economic growth situation. Given that monthly data and quarterly data for China's marine economic indicators are not published,annual low frequency marine economic data analysis in a short sample data is not accurate. Here,we constructed a MF-VAR model to measure China's marine growth and using real-time data from 2005 to 2013 optimised selection and parameter estimation for the MS-VAR model. We then compared the results of the benchmark model to explore the application of the frequency mixing data model to the marine economy. The result shows that MF-VAR modeling has relatively smaller error when applied to China's measurement of marine growth,and the fitting effect of the multivariable MF-VAR model is better than the single variable MF-VAR model of GOP-CIFA and GOP-VFH separately. This illustrates that cyclic fluctuation in the marine economy is affected by various factors and only the influence degree is different. Compared with the benchmark model,the MF-VAR model has a comparative advantage in the accuracy of short-term forecasting. Estimation and prediction precision decrease with increasing forward prediction steps. The MSE of both univariate and multivariate MF-VAR models are lower than corresponding data with the same frequency mixing benchmark model. The frequency mixing data model can not only solve the problem of short sample length,but also has advantages of extracting high frequency data for the marine economy and improving economic analysis accuracy and timeliness.

Key words: benchmark model, marine economy, MF-VAR model, mixed frequency data