基于ARMA-GARCH组合模型的汇率波动性预测

Exchange Rate Volatility Forecasting Based on ARMA-GARCH Combined Model

文章使用2015年8月11日至2022年11月30日人民币兑美元汇率的日交易中间价数据进行研究。首先对该数据进行对数差分处理,得到平稳的人民币汇率对数收益率序列。其次通过描述性统计分析发现该序列“尖峰厚尾”,而平稳性检验和ARMA模型建立证实了该序列的非随机性质,异方差性检验确认了波动聚集性。研究采用GARCH、EGARCH和TGARCH模型对波动率方程进行拟合,结果显示汇率存在杠杆效应。最终,使用最佳拟合效果的EGARCH(1,2)模型对汇率序列进行回测分析,并提出相关建议。

This paper uses daily transaction middle price data of the RMB-USD exchange rate between August 11,2015 and November 30,2022 for research.Firstly,the data is processed by logarithmic difference to obtain a stable logarithmic return rate sequence of the RMB exchange rate.Secondly,through descriptive statistical analysis,it is found that the sequence has a“peaked ness and fat tail”.The stationarity test and ARMA model establishment further confirm the non-randomness of the sequence,and the heteroscedasticity test confirms the volatility clustering.This study uses GARCH,EGARCH and TGARCH models to fit the volatility rate equation,and the results show that there is a leverage effect in the exchange rate.Finally,the EGARCH(1,2)model with the best fitting effect is used for backtesting analysis of the exchange rate sequence,and relevant suggestions are proposed.