基于集合经验模态分解的上证指数波动特性分析

发布时间：2018-11-08 20:37

【摘要】：我国股票市场经过20多年的发展已经成为我国市场经济的重要组成部分，股票市场的健康发展是我国经济稳定发展的重要基础。经济学界的众多学者认为，股票市场的价格波动与宏观经济之间的关系密切。对两个市场间的关系进行深入地理论研究和实证检验，有利于资本市场的完善和健全，有助于股票市场的发展，对推进我国社会主义经济建设和市场化改革起到至关重要的作用。因此，近年来经济金融学界越来越关注对股票市场价格波动与宏观经济波动之间相互关系以及作用机理的研究。美国工程院士黄锷博士于1998年提出了一种新的信号分解方法——经验模态分解算法（Empirical mode decomposition，EMD），这种方法的本质是通过数据的时间尺度特征来获得本征波动模式，然后分解数据。依据数据自身的时间尺度特征来进行信号分解，而无须预先设定任何基函数。与建立在先验性假设的谐波基函数（或基频）和小波基函数上的傅里叶分解与小波分解方法相比，它可以更准确地反应系统原始的物理特性，有更强的局部表现能力，在处理非线性、非平稳信号或时间序列时往往更加有效。集合经验模态分解（Ensemble Empirical modedecomposition, EEMD）方法基于EMD算法，通过加入白噪声的方式，解决了EMD方法中的模式混淆问题，，能够更精确的分解数据序列。本文首先利用信号分解中的EEMD方法，对上证指数的价格序列进行分解，并通过平均周期、方差贡献、Pearson相关系数三个指标的角度对其进行评价。进一步地，将分解得到的IMFs划分为高频、低频、趋势三个分量，并对三个分量的经济学意义进行解释。本文认为趋势分量代表了上证指数的长期运行趋势，在本文的考察期内上证指数的长期趋势是上升的，但上升趋势有所减缓；低频分量反映了上证指数的中长期波动，其波动方向与上证指数基本一致，本文认为低频分量的波动受到宏观经济的影响，并从直观上发现低频分量中的IMF6、IMF5分别与CPI和工业增加值的走势十分接近；高频分量的平均周期短，代表了上证指数的短期波动，其主要受到市场波动如投资者心理、短期事件政策刺激、外围市场的波动溢出效应等因素影响。根据分解出的不同频域分量的基本分析，本文对低频和高频分量有针对性的分别建立相应模型进行分析。针对低频分量，本文将代表上证指数中长期波动的低频分量单独提取出来，与宏观经济变量中的通货膨胀率、工业增加值、货币供应量M1三个变量建立VAR（3）模型，发现四者之间存在着长期均衡关系。通过协整检验和方差分解，得出结论认为通货膨胀与低频分量负相关，即通货膨胀会遏制上证指数中长期的向上波动趋势，而工业增加值的增长能够刺激上证指数的中长期上涨。同时，通货膨胀与经济增长对上证指数中长期趋势的影响程度相当，而货币供应量的增长虽然也有利于股市的中长期上涨，但其影响程度较低，表现为长期的“货币中性”。针对高频分量，本文主要分析上证指数与不同地区股票市场高频波动间的相互影响。通过EEMD方法将股票指数的高频分量提取出来，单独研究他们之间的相关性，更能够说明不同市场波动间的短期影响。本文对上证指数、香港恒生指数、美国道琼斯工业指数的高频分量分两阶段进行Granger因果分析，发现第一阶段（1997年至2006年）上证指数的高频波动与恒生指数和道琼斯工业指数相关性很弱，表现出一定的独立性；而在第二阶段（2007年至2014年）上证指数与道琼斯工业指数高频波动间存在着双向的格兰杰因果关系，相比于第一阶段，在第二阶段两者的短期波动之间存在着显著的相互影响。通过分析认为其主要原因在于我国汇率改革进程、资本账户的开放程度不断加深，QFII与QDII的迅速发展是中美两国资本市场间联系程度加深的主要原因，另外由于美国在国际金融体系中的地位，使其次贷危机对资本市场的影响更容易传播到其他市场；而第二阶段港股的短期波动对上证指数有一定影响，相对于道琼斯工业指数，上证指数对恒生指数的短期波动影响有限。总之，本文将EEMD方法与传统计量方法结合，通过EEMD分解方法将上证指数价格序列分为三个频率层次的波动，分别研究不同频率波动的影响因素，有助于准确全面的分析不同因素对上证指数波动的影响，为投资者在股市中追求最大化收益提供建议，还能为决策层防范金融风险，制定金融决策提供依据。
[Abstract]:The stock market has become an important part of China's market economy through the development of over 20 years, and the healthy development of the stock market is an important foundation of the development of our country's economy. Many scholars in the economic circle believe that the price fluctuation of the stock market is closely related to the macro-economy. The relationship between the two markets is in-depth theoretical research and empirical test, which is beneficial to the improvement and the sound of the capital market, which is conducive to the development of the stock market and plays an important role in the promotion of the socialist economic construction and the market-oriented reform of our country. Therefore, in recent years, more and more attention has been paid to the relationship between the price fluctuation of the stock market and the macro-economic fluctuation and the study of the mechanism of action. In 1998, a new method of signal decomposition _ empirical mode decomposition (EMD) was proposed by the American Academy of Engineering. data. The signal decomposition is performed according to the time scale characteristics of the data itself, without the need to set any base in advance. Compared with the wavelet decomposition method, the harmonic basis function (or fundamental frequency) and the wavelet-based function of a priori hypothesis can more accurately reflect the original physical characteristics of the system, have stronger local performance, a linear, non-stationary signal, or time sequence, often more The ensemble empirical mode decomposition (EEMD) method, based on the EMD algorithm, solves the model confusion problem in the EMD method by adding white noise, and can more accurately decompose the data. In this paper, the method of EEMD in signal decomposition is used to decompose the price sequence of the index, and the angle of the three indexes of Pearson's correlation coefficient is calculated by mean period, variance contribution and Pearson correlation coefficient. further, the decomposed IMFs are divided into three components of high frequency, low frequency and trend, and the economic meaning of the three components is This paper thinks that the trend component represents the long-term running trend of the Shanghai stock index, and the long-term trend of the index in the period of the investigation is up, but the rising trend is slow. The low-frequency component reflects the medium and long-term fluctuation of the Shanghai stock index, and its direction of fluctuation and the above-mentioned index refer to the above-mentioned index. The results show that the fluctuation of the low-frequency component is influenced by the macro-economy, and the IMF6 and IMF5 in the low-frequency component are found to be close to the trend of the CPI and the industrial added value, respectively. The average period of the high-frequency components is short, which represents the Shanghai Stock Index. The short-term fluctuation of the market is mainly affected by market volatility, such as investor psychology, short-term event policy stimulus, volatility spillover effect of peripheral market On the basis of the basic analysis of the decomposed different frequency-domain components, this paper sets up the corresponding establishment of the low-frequency and high-frequency components, respectively. Based on the low-frequency component, the low-frequency component, which represents the medium and long-term fluctuation of the index index, is extracted separately, and the VAR (3) model is set up with three variables of the inflation rate, the industrial added value and the money supply quantity M1 in the macro-economic variable. In the long-term equilibrium relationship, it is concluded that the inflation is negatively correlated with the low-frequency component, i.e., the inflation will contain the long-term upward fluctuation trend of the Shanghai stock index, while the growth of the industrial added value can stimulate the upper card. In the medium and long term of the index, the effect of inflation and economic growth on the long-term trend in the Shanghai Stock Index is comparable, and the growth of money supply is also beneficial to the mid-and long-term increase of the stock market, but its impact is low, and it appears as a long-term "water" currency y ". For high frequency components, this paper mainly analyzes the Shanghai stock index and the stock market in different regions The high-frequency component of the stock index is extracted by the EEMD method, the correlation between them is studied separately, In this paper, the high-frequency components of the Shanghai Stock Index, the Hong Kong Hang Seng Index and the Dow Jones Industrial Index were analyzed in two stages, and the high-frequency fluctuation and the Hang Seng index and the track of the first stage (1997-2006) were found. The Jones industrial index has a weak correlation and has a certain independence; in the second stage (2007-2014), there is a two-way Granger causality between the index and the high-frequency fluctuation of the Dow Jones Industrial Index, compared with the first phase, the short-term fluctuations in both the second and the second phase The main reason of the analysis is that China's exchange rate reform process and the opening degree of capital account are deepening, and the rapid development of QFII and QDII is the main reason for deepening the relationship between the two countries' capital markets. In the international financial system, the impact of the second loan crisis on the capital market is more likely to spread to other markets, while the short-term volatility of the second stage of the Hong Kong shares has a certain impact on the Shanghai Stock Index, with respect to the Dow Jones Industrial Index, the Shanghai Stock Index and the Hang Seng Index. In summary, the EEMD method is combined with the traditional measurement method, and the price sequence of the Shanghai stock index is divided into three frequency-level fluctuation by the EEMD decomposition method, and the influence factors of different frequency fluctuation are studied respectively, so that the accurate and comprehensive analysis is facilitated. The influence of different factors on the fluctuation of the Shanghai stock index is to provide advice to the investors in the stock market for maximizing the income, and also to guard against the gold at the decision-making level.
【学位授予单位】：浙江财经大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：F832.51;F224

【参考文献】