Research On The Relevance Between Capital Market And Real Industry

Under the framework of portfolio theory and efficient market theory,there is a strong relationship between capital market and real economy.The pricing basis of capital market is the real economy,and its value fluctuation is decided by the real economy.The early economists mainly studied the influence of the real economy on the value of the capital market.In recent years,with the development of capital market,rationalization of market participants and improvement of the computer,more and more economists are digging the capital market to research the real economy.The capital market as a macroeconomic indicator is strengthened.Traditional finance focuses on the relationship between capital market and macro economy and also individual company.As a whole,the capital market can reflect the current situation and expectation of the macro economy.While as a single stock,the capital market can react to the listed company's operating status.In recent years,the global macroeconomic downward,traditional industries in the manufacturing industry and primary processing industry mainly gradually decline,while the emerging industry in the Internet and information technology is mainly to super normal development,industry in the macro economy appeared serious differentiation,so the research of the industry has gradually become a research hotspot.China has become the most important stabilizer of the world economy since the beginning of the 21 st century,especially the developed countries experienced a series of crises,such as the financial crisis in 2008 and the European debt crisis.However,under the background of maintaining rapid economic growth in China,structural problems become more and more prominent.Structural imbalance directly reduces the quality and economic efficiency of economic development,increases environmental costs,which leads social problems increasingly serious.The sustainable economic development is threatened.As one of the most important and largest economies in the world,the imbalance of industrial structure needs to be solved urgently.As a new hotspot,complex networks have attracted the interest of researchers and young scholars from different disciplines.China has held a number of academic conferences and forums on the theme of complex networks.In recent years,more and more scholars and researchers have combined complex networks with industrial research to study the structural characteristics of industrial networks.Complex networks focus on analyzing the intertwined relationship between nodes.The analysis method of complex networks can effectively depict the relevance between the nodes.Based on the financial transaction data of the capital markets,this paper constructs the industrial complex network,and obtains the results of industrial transition by the comparative static method.With financial transaction data based on capital markets,this paper mainly studies from the following aspects:(1)The Statistical Methods in the Process of Industrial Complex Networks ConstructionIn this paper,the main core of statistical methodology is complex network,and the specific detail of the process is the construction of industrial networks.This paper uses the parameters of the FF three-factor model as the attributes of individual companies.Based on these parameters,this paper gets the sub-industries in the industry through clustering.During the process,it is necessary to normalize the data,including the same dimension processing and the same influence processing,so as to ensure the three parameters of the cluster with the same important role,which makes the clustering result reliable.This paper creatively proposes Adaptive-CFSFDP clustering method,which can automatically obtain the clustering density cores and the number of clusters by the best CDbw index,the clustering effect evaluation index.For the edge weights of industrial networks,this paper uses the mutual information entropy to measure,to represent the intertwined relationship between the two node variables.(2)The Construction Process of Industrial Complex NetworksThe construction process of industrial complex network is one of the core of this paper.A complete network includes nodes,edges and specific network model.In this paper,the industrial network adopts the Adaptive-CFSFDP clustering method to aggregate the listed companies in the same broad categories of industries to obtain the industry nodes of the subindustry network.The mutual information entropy of the index yield series formed by the subsectors is taken as the edge of the network,and select the fully coupled undirected weighted network as the industrial network model.We have formed a large-scale industrial network model by integrating the sub-sectors.Therefore,we obtain two industrial complex network models: the sub-industry network model and the industry integrated network model.(3)The Result Analysis of Industrial Complex NetworksBased on the weekly earnings data of listed companies from 2005 to 2015,the index is formed by the sub-industry nodes clustered with listed companies,which weighted by the market value.At the end of each month,the three-year time series data of return is used to calculate the mutual information entropy as the edge weights,and then forms a sub-industry network.We can obtain an integrated industrial network through the synthesis of sub-sectors.By studying the total number of sub-sectors in all industries and the similarity of the community structure formed by the adjacent networks in the same industry,the results of the changes within the industry are obtained.By comparing the topology and static characteristics of the network at different time,the result of the change between industries can be obtained by the method of comparative static analysis.And thus the formation of the basic facts of industrial change: changes within and with the industries.(4)Sensitivity Analysis of Industrial Complex Network ResultsIn order to demonstrate the validity of the method and the stability of the results,we introduce the minimum connected network,the minimum spanning tree and the network invulnerability method.We use Spearman Rank Correlation Analysis and the Similarity Analysis of Leading Industries between the four industrial transitions results based on nodes strength method and the other dynamic methods,so as to observe whether the same conclusion can be obtained by using different network analysis methods.In addition to the sensitivity analysis of the method,the sensitivity analysis of parameters is also very important in the empirical analysis.Although this paper emphasizes the methodology for the use of parameters is less,because of the characteristics of the method,especially the network correlation relationship-mutual information entropy calculation process to use a number of parameters,so we mainly focus on the mutual information entropy calculation process parameters Sensitivity analysis,including the calculation method,the frequency of data used and the calculation of the cycle.With all the above research processes,this paper mainly draws the following conclusions:(1)It is necessary to standardize the data.While the data of the social discipline seems to have the same dimensions and data type,but we can find the dimensions and influence to the stock yields of the three parameters of FF three-factors model are different after careful analysis.The study of the construction process of the three factors can be found that the volatility of the three factors is different,that is,different dimensions,the need for the same volatility,and also for the same influence.There is a strong correspondence between the clustering results after data processing and the classification of the real industry.We modify the classical CFSFDP clustering method in detail,and then propose a new clustering method Adaptive-CFSFDP.By picking the best CDbw index,we can obtain the clustering cores and results automatically.We apply this method to several classical clustering analysis cases and get the expected results.(2)According to the above-mentioned network construction process,we get the globalcoupled undirected weighted network using the 2005-2015 data of listed companies.By the comparative static analysis method,we find the fact of industrial transition,including the transition of inter-industry and the industries.From the analysis results,we can see that inter-industry transition really exist,where the numbers of segments within the industry,or the community structure of the company within the industry.The industries may be merged,split,demise,newborn or other situations with national economic development stage.It is precisely because of this change within the industry that make the industry maintained vitality and competition,which could deal with the impact of other industries,extend the industry's life cycle as far as possible.For the changes between industries,we have found the following facts: in the ten years of statistical range,the leading industry has undergone a trend of transfer,information technology and education industries gradually replace the traditional industries to become the leading industry,also benefit from the Internet and the emerging network consumption tide delivery warehouse industry steadily in recent years,has entered the leading industry series.National policy led to some industries in the stage to become the leading industry,the performance of the second half of 2011 and the first half of 2012 the real estate industry for several months ahead of the previous.Some industries are perennial ranking ahead,maintaining a good economic vitality.Some industries are insufficient demand and supply surplus,perennial ranking later,but with the national supply side of the reform policy in depth,mining and other industries supply and demand situation is expected to be advanced stage.Some industry supply and demand stability,perennial ranking in the middle(3)This paper compares the method based on the comprehensive ranking of nodes strength and the minimum connected network,the minimum spanning tree and the network invulnerability.We find that the rank correlation coefficient of all rankings is above 0.8.The similarity of leading industries is more than 0.93,which indicated the validity of the method and the reliability of the results analysis.In the aspect of parameter sensitivity,we analyze the calculation method of mutual information entropy,data frequency and calculation period.The results show that the calculation method of mutual information entropy and the data of daily,weekly or monthly data have little influence on the results,and the stability is good.The results show that the stability of the results is better when the time is more than two years,while when the data cycle is used for one year,the results are weak.Through statistical analysis,we found that using one year data,because the data quantity is less,the probability distribution of variables description is not comprehensive enough,the probability of mutual information entropy distribution is non normal characteristics.Therefore,the correlation is not good,especially in the year of sharp fluctuations.Therefore,in order to study the industry network in the mid-term industrial transition,we need to be alert to the impact of the data cycle,and use the appropriate data to analyze.The theoretical principle of the capital market and the real industry relevance is the effective market theory and portfolio theory,while this paper is the empirical evidence.It adopts the dynamic analysis method of complex network,and has some inspiration and reference to the further research.