Abstract: To solve the higher peak and fat tail phenomenon，immediate memory and asymmetric features，this paper formulate the volatility model of exchange rate returns using the ARFIMA-EGARCH-M model.Based on the example on the return sequences of CNY/EUR exchange rate，assuming the error sequence is Normal distribution，T distribution，GED distribution and SKT distribution respectively，authors derive that the ARFIMA（1，d，1）-EGARCH（2，2）-M with GED distribution can better solve the higher peak and fat tail，immediate memory and asymmetric features.

Key words: ARFIMA-EGARCH-M model, higher peak and fat tail, immediate memory, asymmetric characterization

摘要： 为了解决汇率收益率波动中的“尖峰厚尾”、中期记忆和非对称特征，提出了利用ARFIMA-EGARCH-M模型建立汇率收益率波动模型。以人民币/欧元的日汇率数据为例，在误差分布分别为正态分布、T分布、GED分布与SKT分布的前提下分别进行模型拟合分析，得出基于GED分布的ARFIMA（1，d，1）-EGARCH（2，2）-M模型拟合人民币/欧元汇率收益率效果最好，能较好地解决“尖峰厚尾”、中期记忆和非对称现象。

关键词: ARFIMA-EGARCH-M模型, 尖峰厚尾, 中期记忆, 非对称