摘要
近几年来,伴随人民币汇率体制改革以及升值压力的增加,人民币汇率波动行为也日益引起国内外广泛的关注。越来越多的研究表明金融数据存在着非线性行为,以人民币汇率时序的非线性和复杂性为主要研究对象,通过BDS检验从统计推断角度验证了其非线性的存在,从而建立自激励门限自回归模型来拟合数据,同时采用GARCH模型消除残差异方差现象,在理论和实践上都有较大的改进和尝试,拟合结果显示预测平均相对误差为0.0694%。通过对人民币汇率非线性特征的研究,不仅为其管理提供了支撑,同时对非线性时间序列问题的进一步讨论提供了理论和实证的依据。
In recent years,along with the reform of RMB exchange rate system and appreciation pressure,the RMB exchange rate fluctuation has increasingly attracted widespread attention at home and abroad.Given that a growing body of research shows that financial data is charactered with non-linear feature,the text sets the non-linear and complex features of the RMB exchange rate time series as the main object of study.Firstly,BDS test verifies the presence of non-linearity from the perspective of statistical inference,on the basis of which the text makes the self-excitation threshold autoregressive model to fit the data,at the same time,GARCH model is brought in to eliminate the heteroscedasticity.Comparing the fitted data with the actual data,it shows that the average relative error is 0.0694 percent.The non-linearity study of RMB exchange rate not only makes contributions to government management,also provides a theoretical and empirical basis for further discussion of the non-linear time series.
出处
《中南财经政法大学研究生学报》
2013年第1期44-52,共9页
Journal of the Postgraduate of Zhongnan University of Economics and Law