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 ›› 2026, Vol. 34 ›› Issue (3): 39-50.doi: 10.16381/j.cnki.issn1003-207x.2022.0618cstr: 32146.14.j.cnki.issn1003-207x.2022.0618

• • 上一篇    下一篇

基于事件流模型的粮食期货极端风险测度计算研究

蒋志强1, 胡海燕1, 戴鹏飞2(), 王莉1, 周炜星1   

  1. 1.华东理工大学商学院,上海 200237
    2.武汉理工大学管理学院,湖北 武汉 430070
  • 收稿日期:2022-03-29 修回日期:2022-07-02 出版日期:2026-03-25 发布日期:2026-03-06
  • 通讯作者: 戴鹏飞 E-mail:pfdai@whut.edu.cn
  • 基金资助:
    国家自然科学基金资助项目(U1811462);国家自然科学基金资助项目(72171083);国家自然科学基金资助项目(72201099);中央高校基本科研业务费专项资金项目(104972025RSCbs0070)

Estimating Extreme Risk Measures of Grain Future Market Based on Event Flow Models

Zhiqiang Jiang1, Haiyan Hu1, Pengfei Dai2(), Li Wang1, Weixing Zhou1   

  1. 1.School of Business,East China University of Science and Technology,Shanghai 200237,China
    2.School of Management,Wuhan University of Technology,Wuhan 430070,China
  • Received:2022-03-29 Revised:2022-07-02 Online:2026-03-25 Published:2026-03-06
  • Contact: Pengfei Dai E-mail:pfdai@whut.edu.cn

摘要:

将粮食价格维持在合理范围,防止极端波动发生是保障国家粮食安全的首要任务。本文以粮食期货价格为研究对象,聚焦粮食期货价格极端波动,运用事件流模型对极端波动的发生规模和发生过程同时建模,并应用于极端风险测度的计算。针对中美粮食期货(CBOT稻谷、CBOT小麦、CBOT玉米、郑商早籼、郑商强麦、大商玉米)价格的研究表明,事件流模型可以较好地刻画粮食期货价格波动的胖尾特征和集聚特征,对极端波动的发生规律和发生规模表现出较好的拟合效果。针对极端风险测度的样本内外检验表明,事件流模型可有效地提升风险测度计算的准确性。本文的研究结果不仅有助于更好地理解粮食期货价格极端波动的发生模式,也为粮食价格风险管理和粮食市场平稳运行提供了科学思路和技术手段。

关键词: 极端风险, 极值理论, 自回归条件久期模型, 霍克斯过程, 在险价值

Abstract:

Food plays a critical role in human survival and development, and has always been the global focus. Recently, natural and social events such as rising energy prices, frequent extreme weather, the severe COVID-19, and the Russian-Ukrainian conflict have jointly pushed up the food prices. Keeping food price staying in a reasonable range and preventing extreme fluctuations are crucial to ensure national food security. It is of strategic significance to pay attention to the stable operation of the grain futures market, extreme risk management and early warning.By utilizing the prices of grain futures in the US (CBOT rice, CBOT wheat, CBOT corn) and the Chinese markets (ZCE rice, ZCE wheat, DCE corn), the data are described and visualized, and the clustering characteristics of extreme risks are uncovered.Then event flow models are employed to model the occurring process of extreme fluctuations by means of the ACD-POT model and Hawkes-POT model and derive the formula of extreme risk measures (VaR).Finally, three accuracy test methods are used to evaluate the fitting and prediction effects of the benchmark models (POT, EGARCH-POT), the event flow models (ACD-POT model, Hawkes-POT model), and the ensemble model.The results show that event flow models are able to capture the fat-tailed and clustering characteristics of extreme price fluctuations and fits the price data very well. In addition, further comparison finds that the in-sample and out-of-sample accuracy of Hawkes-POT models is better than other models, and the power-law Hawkes-POT model performs the best. The results not only deepen our understanding of the occurring pattern of extreme price fluctuations in grain futures, but also provide a new avenue for risk management and market supervision in food markets.

Key words: extreme risk, extreme value theory, autoregressive conditional duration (ACD) model, Hawkes process, Value at Risk (VaR)

中图分类号: