The Influential Factors Analysis And The Time Series Analysis Of The Chinese Insurance Industry Development

Insurance,which is as an important part of the financial system,plays an important role to ensure the steady and rapid economic growth and people's living level.Since the domestic insurance business returned to normal in 1979,premium income has increased strongly,which is from 620 million yuan in 1980 to 3.1 trillion yuan in 2016.In the process of rapid development of the insurance,it exposes many problems.Insurance density and depth is low,the regional development imbalances and other issues are also increasingly prominent.Therefore,the influence factors analysis and prediction of the premium income play an important role to strengthen the regulation of insurance and help the company managers to make decisions.When multiple linear model is set up,the correlation between variables may cause multicollinearity problem.This paper applies principal component analysis in panel data model,which can weaken the effect of multicollinearity.Quantile regression model with variable coefficients can analyze the influence of various influence factors on the premium income from various angles in detail.Markov regime switching model can handle data of state transition due to certain factors,and the neural network in dealing with complex data also has a great advantage.This paper selects panel data in 31 provinces and cities over 2007-2014 and establishes a fixed effect variable intercept model by applying principal component analysis to analyse the influence of gross domestic product(GDP),per capita disposable income,total fixed asset investment and other factors.The results of the model show that the influence of the population to premium income is the largest,GDP,total fixed asset investment,urban and rural residents savings are followed,and the influence of the per capita disposable income to premium income is minimal.We propose partial linear variable coefficient quantile regression panel data model and use this model to discuss the influence of various influence factors on the premium income from various angles in detail.The results of the model show that the influence of various factors on the premium income is slightly different under the different premium income levels,and the influence of various factors on the premium income has little difference on the same province at different times.However,the influence of various factors on the premium income has relatively large difference on different provinces at the same time.Then,we set up Markov regime switching autoregressive conditional heteroskedasticity model(MS-ARCH)and BP neural network respectively to analyse monthly premium income time series over 2006-2016.The MS-ARCH model is a good model to analyse the growth rate of premium income volatility,and the BP neural network is used for fitting premium income.By comparing the results,we find that the BP neural network is better than MS-ARCH model in the prediction of premium income,and we use the BP neural network to predict premium income from December 2016 to February 2017.