|
| 基于R/S检验、ARFIMA模型和小波方差的人民币汇率长记忆性检验 |
| Test of long memory characteristics of RMB exchange rate by R/S statistics, ARFIMA model and wavelet variance |
| 投稿时间：2010-11-15 |
| DOI： |
| 中文关键词: 汇率波动 长记忆性 R/S检验 ARFIMA模型 小波方差 |
| 英文关键词: exchange rate fluctuation long memory characteristics R/S statistics ARFIMA model wavelet variance |
| 基金项目:教育部人文社会科学研究项目(09YJC790209)；武汉科技大学基金资助项目(2009xz41). |
|
| 摘要点击次数: 16697 |
| 全文下载次数: 6108 |
| 中文摘要: |
| 分别应用R/S检验、ARFIMA模型和小波方差对人民币兑美元名义汇率收益率序列的长记忆性进行检验。根据R/S统计量计算出Hurst指数为0.5737451,采用ARFIMA(2,d,1)模型对人民币汇率收益率序列进行拟合的效果比较好,其分数差分参数为0.1457,利用Haar小波对人民币汇率波动绝对值收益率序列进行最大重复离散小波变换,得出其长记忆性参数为0.3931。3种方法的研究结果均表明人民币兑美元名义汇率收益率序列存在长记忆性。 |
| 英文摘要: |
| This paper determines the long memory characteristics of the USD/RMB nominal exchange rate by R/S statistics, ARFIMA model and wavelet variance. According to the R/S statistics, the Hurst index is calculated to be 0.5737451. The result shows that the fitting effect of ARFIMA (2,d,1) model is better and the parameter of fractional difference is 0.1457. MODWT (maximal overlap discrete wavelet transform) is carried out of absolute returns series of exchange rate fluctuations by Harr wavelet and the long memory parameter is obtained to be 0.3931. The results of three methods all show that the USD/RMB nominal exchange rate has long memory. |
|
查看全文
查看/发表评论 下载PDF阅读器 |
| 关闭 |
