Pricing of life insurance products using markovian aging process model

In this research, our main objective is more accurate pricing of life insurance products with a new approach of predicting mortality or survival rates. Currently, a life table is used to calculate the current value of pensions, insurance premiums, etc. Therefore, to increase the accuracy of our calculations, we are looking for a mortality prediction model for such calculations. Therefore, in this research, instead of static pricing (only using the latest life table), we used life table prediction and dynamically rated life insurance products.

In this research, a new model proposed to predict the probability of human mortality (survival) based on the Markov process, a limited state with an absorption state (death). This model measured based on the physiological age, because the physiological age of each person can be checked based on different laboratory indicators, and finally it has led to the results of the individual health index. In addition, the parameters of this model are the initial probability vector and the sub-intensity matrix of a Markov chain that changes over time. In other words, in this model, according to a possible process in the model, the initial probability vector over time selects the possible interval of the physiological age equivalent to the chronological age.

To show the satisfactory performance of this model, the relevant data set from the United States of America was analyzed. The predicted results with the presented model are better than Lee Carter''''s model. It should be noted that the number of parameters of the model introduced in this research is much less compared to the Lee Carter model and other mortality or survival prediction models. Based on this model, a closed form for life insurance pricing relationships is obtained, which simplifies these calculations for users.

The relationships obtained for pricing were investigated based on two products, 5-year term life insurance and also a 5-year term pension. The fitted results for the model used in the predictions of the probability of mortality as well as the probability of survival and pricing are very satisfactory.