摘要
针对Brent原油期货市场可能存在结构突变点(结构断点),引入了隐马尔科夫模型(HMM)对其进行波动状态预测,但是由于HMM模型测度下的波动状态中可能存在伪结构突变点,再使用迭代累积平方和(ICSS)模型对Brent原油期货市场波动结构突变点进行诊断,并修正其波动状态。为了检验ICSS-HMM-EGARCH模型对Brent原油期货市场结构突变点预测的准确性,基于修正后的波动状态再次使用HMM-EGARCH模型对Brent原油期货市场进行波动率预测。最后,采用成功率(SR)和基于平均误差函数(RMSE)的Diebold-Mariano(D-M)模型分别对预测波动状态和预测波动率的准确性进行检验。实证结果表明:Brent原油期货市场中存在波动结构突变点;HMM模型测度下的波动状态中存在伪结构突变点,而ICSS-HMM-EGARCH模型能够修正波动状态中的伪结构突变点;基于修正后的波动状态后HMM-EGARCH模型能够对Brent原油期货市场进行更加准确的波动率预测,因而ICSS-HMM-EGARCH模型能够准确地预测Brent原油期货市场波动结构突变点。
There may exist a structure breakthrough point in Brent crude oil futures market due to the impact of financial crisis. In this paper, the hidden Markov model is introduced to forecast the volatility state in Brent crude oil futures market. However, there may exist some supposititious volatility structure breakthrough points based on the HMM model. Therefore, an ICSS-HMM-EGARCH model is built to re-forecast structure breakthrough points in Brent crude oil futures market based on iterated cumulative sums of squares, and the volatility state is then corrected. In order to test the reliability of forecasting structure breakthrough points by using the ICSS-HMM-EGARCH model, the volatility of Brent crude oil futures market is re-forecasted based on the corrected volatility state once again. Finally, the success rate is used to evaluate the forecasted volatility state, and the forecasted volatility is tested with the Diebold-Mariano model based on the root mean squared errors function. The empirical results show that there exist structure breakthrough points in Brent crude oil futures market and there exist supposititious volatility structure breakthrough points based on the HMM model, but the ICSS-HMM-EGARCH model can correct the supposititious structure breakthrough points effectively. The HMM-EGARCH model can obtain a more accurate volatility based on corrected volatility state. Therefore, the ICSS-HMM-EGARCH model is capable of effectively forecasting the structure breakthrough points in Brent crude oil futures market.
作者
陈粘
林宇
黄登仕
陈宴祥
CHEN Zhan;LIN Yu;HUANG Dengshi;CHEN Yanxiang(College of Management Science,Chengdu University of Technology,Chengdu 610059,China;School of Business,Chengdu University of Technology,Chengdu 610059,China;School of Economics and Management,Southwest Jiaotong University,Chengdu 610031,China)
出处
《系统管理学报》
CSSCI
CSCD
北大核心
2019年第6期1095-1105,共11页
Journal of Systems & Management
基金
国家自然科学基金资助项目(71771032)
国家社会科学基金资助项目(17BJY188)
四川省应用基础研究资助项目(2017JY0158)