Spatial Statistical Analysis Of Listed Companies In China

Recent years have seen a great development in spatial statistics while there isn't much research in listed companies.However,locations of listed companies are very important information.Therefore as a significant tool for detecting effects of geographic information,it is necessary to make use of spatial econometrics in the area of financial research about listed companies and their stock yield.Firstly,the author makes a brief review of the basic theory in related spatial statistics.According to the former researches,the author describes the descriptive statistics of listed companies by different information including geographic information,industry information and financial document information and then represents the results on the maps.The author also tries to mix the spatial weight matrix that reflects geographical distance relationships between various companies with financial document information into a new distance matrix for listed companies clustering analysis.The results imply that listed firms aggregate according to their geographical location especially in Beijing area,Shanghai area and Shenzhen area which are developed areas in China.It also shows that the distribution of listed companies is geographical unbalanced not only in amounts but also in quality.Secondly,the author exploits spatial weight matrix and industry matrix of listed companies to build spatial lag model in order to quantify spatial autocorrelation.The research shows the stock yields of different listed companies are more likely to have the trends of same rise and fall if they are geographical adjacent and belongs to the same industry.Moreover,the influence of industry is more apparent then geographical distance.Finally,the author makes a try to extend spatial panel model in dynamic ways by adding time lag in.Based on the basic knowledge of dynamic panel model and spatial lag model,the author makes further research of model specification,determination of lag order,selection of spatial weight matrix and model strengths.Besides by comparing fitting effects of various kinds of models,it comes to the conclusion that dynamic spatial panel model has advantages over spatial lag model and dynamic panel model.Moreover,spatial lag model with time lag is best-fitted model in the data of listed companies' monthly stock yield among all the other models and has a great forecast efficiency generally but differed slightly in diverse companies.