基于EMD-WA模型的BDI指数波动周期特征研究

运用结合EMD分解和小波分析构建的EMD-WA模型,分析了国际干散货市场BDI指数的波动周期特征.通过EMD方法对BDI指数序列进行分解,针对分解出的本征模函数的均值、方差比和周期三个方面的特点,将不同频率的IMF进行分类重构,最后通过小波分析方法得到重构序列的主周期情况,并对结果做进一步分析.研究结果表明:一是BDI序列可以分解为随机波动项、周期波动项和趋势项三部分:二是季节性波动周期长度约为1年(0.88年),短周期长度约为4年(3.70年),中周期长度约为10年(9.96年);三是BDI序列周期特别是短周期呈明显4年到2年的缩短趋势;四是突出分析了重大事件与中、长周期紧密联系.

Study on the periodic characteristics of BDI index based on EMD-WA model

Using the EMD-WA model constructed by EMD decomposition and wavelet analysis, this paper analyzes the fluctuation cycle characteristics of BDI index in international dry bulk market. EMD method is used to decompose the BDI index sequence, in view of the three aspects of the mean, variance ratio and period of the decomposed eigenmode function, the IMF of different frequencies are classified and reconstructed. The main cycle of the reconstruction sequence is obtained by wavelet analysis, and further analysis is made on the result. This research indicates that: First, the BDI index can be decomposed into three parts: stochastic volatility, periodic fluctuation and trend. Secondly, the periodicity of the seasonal fluctuation term is about 1 years (0.88 years), the cycle length of short period fluctuation term is about 4 years (3.7 years), the cycle length of the median period fluctuation term is about 10 years (9.96 years). Thirdly, the BDI index cycle, especially the short period, showed a significant shortening trend of 4 to 2 years. Fourthly, it analyzes the close relationship between the major events and the middle periods or the major events and the long periods.