An Empirical Study On Price Fluctuation Of China's Stock Market Based On Linkage And High-frequency Time-varying

Since the second half of 2014,the variance of stock market volatility comparable to the financial crisis in 2008,in such a volatile situation,a more comprehensive study of the characteristics of volatility is particularly important for investors and regulators to provide short-term fluctuations useful information.This paper focuses on the characteristics of stock market volatility,and analyzes the characteristics of volatility into micro-level linkage characteristics and macro-level volatility characteristics based on high-frequency time-varying,aiming to study the volatility characteristics more deeply and comprehensively.In this study,we use the cluster analysis method to analyze the structure of the internal constituents of the Shanghai 50 Index,aiming to compare and analyze the internal linkage characteristics of the index during different periods.First,the daily closing price data of the 47 constituent stocks after screening are processed.The processed sequence is defined as the revenue-volume series,and the sequence is symbolized according to the quartile to obtain a new symbolized sequence.According to this new symbolized sequence,clustering is carried out from four different fluctuation characteristic stages.From the perspective of the change of the number of clusters in four stages and the bifurcation of the minimum spanning tree,the linkage of the industry is gradually weakened with the fluctuation,and can not return to the fluctuation immediately after experiencing severe fluctuation Before the structural features.In the industry linkage,the industry's financial industry linkage is the most stable,reflecting the index structure when the financial sector accounted for a larger rationality.In addition,there are other clustering features such as locality and auditor institution similarity.Based on the empirical mode decomposition(EMD)method proposed by Huang(1998),this paper firstly introduces the ensemble empirical mode decomposition(EEMD)signal decomposition method based on the EMD,using EEMD method to get the high frequency components.The signal sequences are decomposed by adding white noise,then the high frequency components are extracted and reconstructed.As the extracted high-frequency components have noise properties,a series of tests are carried out.The ARMA(3,2)-GARCH(1,2)model was used to regress the extracted high frequency components.In the regression analysis,we first analyze the high-frequency components that do not consider the volume increment factor.We find that the influence of past volatility on the current volatility is negative before the trading volume is taken into account.That is,past volatility will reduce the current volatility.Secondly,we analyze the high-frequency components which take into account the increment of volume.We find that the past volatility of the index has become positive for the current volatility,that is,the past volatility will increase the current volatility.Through the above empirical analysis,and from the perspective of behavioral finance,draw the following conclusions:(1)industry linkage in the index fluctuations in the dominant,especially the linkage of the financial sector is the most stable,reflecting the index structure(3)the volatility of the index in the short term because of the disclosure of information,investor sentiment,etc.(3)the volatility of the index will fluctuate in the short term;The influence of factors on future volatility in different directions reflects not only the existence of irrational phenomena such as "herding effect" in the stock market,but also the characteristics of "information stock market";(4)both the linkage change and the index Short-term volatility characteristics,all reflect the stock market volatility "memory."Finally,some suggestions are put forward from the angle of investors and supervisors.