汇率时间序列非线性特征分析及实证研究

Analysis and Empirical Research on the Nonlinearity of Exchange Rate Time Series

为了更好地描述汇率行为特征,改善对其的短期预测能力,本文分析汇率时间序列的非线性依赖性。通过正态性检验和BDS统计检验,实证研究6种货币兑美元日汇率序列的非线性依赖性;在确定存在非线性依赖的基础上,对子样本序列及标准化序列进行BDS检验,试图确定非线性依赖的来源。结果表明,除欧元外其余5种货币兑美元的汇率序列均存在非线性依赖特征,其形式可能是混沌。

In order to explore the intrinsic characteristics of exchange rates and improve its explanation and forecasting abilities, this paper analyze the nonlinearity dependence lying in the variable. Normal testing and BDS testing were applied to empirical research on six exchange rates among the whole samples. After the discoveries of the nonlinearities lying in the time series, BDS testing was also put on the sub-sample series and the normalized return series, to explore the origin of non- linearity. The results reveal that there exists nonlinear dependence in exchange rate time series of the other five exchange rates except euro and that the nonliterary may be caused by chaos.