主管:中国科学院
主办:中国优选法统筹法与经济数学研究会
   中国科学院科技战略咨询研究院

中国管理科学 ›› 2024, Vol. 32 ›› Issue (1): 13-22.doi: 10.16381/j.cnki.issn1003-207x.2021.0213

• • 上一篇    下一篇

基于马尔科夫和混频数据模型的黄金期货市场波动率预测研究

郭杨莉,马锋()   

  1. 西南交通大学经济管理学院,四川 成都 610031
  • 收稿日期:2021-01-30 修回日期:2022-04-26 出版日期:2024-01-25 发布日期:2024-02-08
  • 通讯作者: 马锋 E-mail:mafeng2016@swjtu.edu.cn
  • 基金资助:
    国家自然科学基金项目(72071162)

Forecasting the Chinese Gold Futures Market Volatility Using Markov-Switching Regime and Mixed Data Sampling Model

Yangli Guo,Feng MA()   

  1. School of Economics and Management,Southwest Jiaotong University,Chengdu 610031,China
  • Received:2021-01-30 Revised:2022-04-26 Online:2024-01-25 Published:2024-02-08
  • Contact: Feng MA E-mail:mafeng2016@swjtu.edu.cn

摘要:

利用马尔科夫机制转换(Markov-switching regime,MS)和混频数据(mixed data sampling,MIDAS)模型,构建新的马尔科夫机制混频模型(MS-MIDAS),并以此模型对中国黄金期货市场波动率建模和预测。运用样本外滚动时间窗(rolling time windows)预测技术和模型信度集(model confidence set,MCS)检验发现:(1)总体上,引入马尔科夫机制转换的混频模型(MS-MIDAS)相比于MIDAS模型本身,从统计上展现出更高的预测精度;(2)含有跳跃的马尔科夫机制混频模型(MS-MIDAS-CJ)的预测精度最高;(3)对不同的预测窗口和不同的滞后阶数(kmax),上述实证结果都是稳健的。

关键词: 黄金期货市场, 波动率预测, 马尔科夫机制转换混频模型, 结构突变

Abstract:

In this study, a new Markov-switching regime (MS-MIDAS) is constructed using Markov-switching regime (MS) and mixed data sampling (MIDAS) models, and the volatility of the Chinese gold futures market is modeled and predicted. Using the out-of-sample rolling window prediction method and the Model Confidence Set (MCS) test, it is found that: (1) In general, higher prediction accuracy is demonstrated by the mixed data sampling models with Markov-switching regime (MS-MIDAS) compared to the MIDAS model; (2) The mixed data sampling model of Markov-switching regime with jumps (MS-MIDAS-CJ) exhibits the highest prediction accuracy; (3) The empirical results remain robust for different prediction windows and different lag orders (kmax).

Key words: gold futures market, volatility forecasting, MS-MIDAS, structural breaks

中图分类号: