Life Insurance Premiums Income Factors Analyze By Co-integration And Forecast By BP Neural Network

Domestic life insurance has been restore in 1982, in the recently twenty years life insurance market has developed with an unprecedented rate, it is one of the domestic fastest growing sunrise industry. Life insurance market has great development potential because it related closely with people's life and daily lives, as the economic developing,financial markets improving and living standards rising constantly. China fulfills its WTO commitments at December 11,2004, from then insurance markets has been fully opened and geographical restrictions of establishment foreign insurance institutions has been canceled, so competition in the domestic life insurance market will be more intense. Insurance premium income is the main source of income for the insurance industry, and has strong influence in management decision, at the same time it can show the life insurance development level in a country or region as a macro indicator. So analysing what are the factors affecting the premium income and the impact and of the factors affecting the premium income and do more accurate prediction of the premium will provide data support for insurance company and China Insurance Regulatory Commission to make planning or formulate policies.China's life insurance is the research object in this paper,and using annual data during 1982 to 2008,combine research at home and abroad, first summarize the main macroeconomic factors have impacting of life insurance, and then from a quantitative point of view using Co-integration techniques analysis long-term relationship and short-term dynamic effects between them.'The results show that the level of economic development, social security, population, the Coefficient of the Old, the rate of urbanization, interest rates have long-term equilibrium relationship and strong one-way granger causality with life insurance premiums, and use this conclusion as the basis of selecting input indicators of neural network. Then established three life insurance premiums BP neural network prediction model with deferent input, and contrast by a series indicator of errors, the results show that when the input index including both the tendency of self-development and the development of relevant factors, the model will have stability predictive and highly forecast accuracy.