Applications Of A Multi-state Risk Factor/Morality Model In Life Insurance

Mortality is decided by three categories of risk factors—the socioeconomic/demo-graphic risk factor,behavioral risk factors and health indicators.Usually this relationship can be reflected in the process of life insurance policy calculation. But now thing is these risk factors in the policy making process is observed,while after that their changes will not be considered.In this thesis,income adequacy, marital status,smoking status,phys-ical activity these four risk factors will be inducted besides the factors of age and gender to define the risk factors of 17 states.First,assume that individuals in the transfer of 17 state follows the non-homogeneous discrete time Markov chain, it can get a given age and gender under the conditions of the 17*17 transition matriices.According to internal data of some insurance company in Shanghai and the Logistic regression model,each step transition probability can be estimated.Considering the health status as a new factor, the model can be expanded to 33 species of state.Under a reasonable assumption,after model updating and recursive calculation it can get 33 kinds of state transition probability of each step.Then replace the fixed constant in the equation of assumption 1 by a sim-ple function whose transition probability is controlled in [0,1] range, and make sure the assumption logical and random.Finally, the discrete-time,multi-state model normally used in the design,pricing and management of life insurance,annuities,pension plans and other financial security programs.It is hoped to help insurance companies to effectively manage the risk of mortality by discussing the effects of extended risk classification, as well as allowing changes of risk factor status.