Public liability of fire risk pricing model and empirical study based on compulsory insurance

With the high speed development of China´s economy, fires occur frequently in public assembly occupancies. Public liability of fire risk is different from that of general liability risk. The generalized linear model (GLM) cannot be used directly in pricing for the public liability of fire risk, whose sample data shows the characteristics of low probability, heavy loss and fat tail. And because of that, there are few persons to study the problem and little relevant literature. The traditional distribution hypothesis of the GLM cannot be met due to the sample has fat tail, this paper uses nonparametric Gaussian kernel replace traditional distribution hypothesis in the GLM model, analyses the Public liability of fire risk, and receives satisfactory results.