The Forecasting Research Of China's Joint-Stock Commercial Banks' Liquidity Risk Based On Neural Network With Particle Swarm Optimization

On the background of global economic integration, China's commercial banks are facing more and more complicated management environment. As one of the important risks that commercial banks are facing with, liquidity risk determines and embodies bank's security and profits of its management. This paper set China's joint-stock commercial banks as investigated objects and did researches about liquidity risk's measurement and forecasting from the static and dynamic angles separately. It also tried to find out the changing rules of banks' liquidity risk on the basis of empirical study on China's four joint-stock commercial banks' liquidity level, so that it can put forward rational proposals to China's commercial banks about how to get rid of liquidity risk and supervise it effectively.The paper firstly summarized domestic and overseas academicians' literatures on commercial banks' liquidity risk. On the study of existing literatures, it ascertained basic methods of commercial banks' liquidity risk's measurement and discussed its essential reasons and influencing factors. It also summed up domestic and foreign states' radical maneuvers about liquidity risk management. Subsequently, it introduced basic definitions and primary assorts on commercial banks' liquidity risk, and described China's joint-stock commercial banks' liquidity actuality, and also did analysis on primary ingredients that contributes to banks' liquidity risk. Then, on the study of main financial risk prediction methods, it introduced a commercial banks' liquidity risk forecasting model that combined Particle Swarm Optimization and Back Propagation Artificial Neural Networks, and chose eight risk metrical indexes from the static angle, it established a rational liquidity index system against China's commercial banks' real liquidity status. The paper set Shanghai Pudong Development Bank as the investigated subject and chose an amount of sample data, then it did some training and testing work on the network model. The testing results showed that PSO-BP arithmetic's forecasted outcome was more accurate that of BP Networks, it was sufficiently close to the real liquidity level. It indicated that PSO-BP arithmetic was a perfect tool predicting liquidity risk. Finally, the article set China's four joint-stock commercial banks(Shanghai Pudong Development Bank, Shenzhen Development Bank,China Merchants Bank and Huaxia Bank),which came to the market more earlier, as the empirical objects, and chose almost fifteen years' quarterly statistical data as research sample, and used the trained network to predict and analyses all the liquidity indexes' changing direction. The empirical results indicated that China's commercial banks' global liquidity level was fine, but it had pressures on short-term and local shortage of liquidity.The article attempted to find new creative method on predicting China's liquidity risk and used a arithmetic, which combined PSO arithmetic and BP networks, to predict banks' short-term liquidity level on the future. It not only promotes the accuracy of liquidity prediction, but also be helpful for the risk administration department to track and monitor the risk. It has guided meanings to improve China's commercial banks' risk management ability.