人民币汇率预测的两种模型  

Two models for RMB/US dollar exchange rate forecasting

作　　者：郭琨[1] 汪寿阳[2] 

机构地区：[1]中国科学院研究生院管理学院,北京100190 [2]中国科学院数学与系统科学研究院,北京100190

出　　处：《系统工程理论与实践》2008年第5期64-69,共6页Systems Engineering-Theory & Practice

摘　　要：通过对汇改后人民币对美元汇率的波动情况进行分析,采用周期-ARMA模型和多变量的CAR模型,对人民币汇率进行短期预测.结果表明,两种模型都能对原始数据达到很好的拟合,对于实际的预测分析,周期-ARMA模型的预测结果较为平稳,而参考了恒生银行汇价的CAR模型对汇价的波动更加敏感.因此,利用两个模型进行组合预测,可以得到更高的预测精度.In this paper, we examine time series features of RMB/US dollar exchange rate, using Cycle-ARMA model and CAR model for short-time exchange rate forecasting. The results show that the two models can both describe the original data well, however, for the actual forecasting, the Cycle-ARMA model has a steady result, but the CAR model which makes reference to the price of Hang Seng Bank is more sensitive to exchange rate fluctuations. Therefore, we can obtain a more precise result using the integration model of the two individual models.

关 键 词：汇价预测 周期-ARMA模型 CAR模型 

分 类 号：F830.92[经济管理—金融学]