Session: 24

Health

* * *

Summary

The aim of the paper is to discuss some problems of the constructions of risk processes models for health insurance. The risk process is an important part of actuarial modelling. For health insurance, it is the most difficult thing to model. If it is believed that some details may be ignored to support the predictive power of the model, a Markov model could be constructed (even in a bit sophisticated way). The Markov property may not be common for health care application. But if we can construct a Markov model it makes the analysis the situation much easier. 

Some examples demonstrate the extremely high degree of stability of the probability of transitions. It is very important for practical use because the actuary or public health planner could pay much more attention to costs of diagnostics and treatment than to risk process estimation. 

Nevertheless, the probabilities of transition may fluctuate. It must be taken into account, but the type of models discussed allows doing it simpler. 

There are at least four directions for further development of the model.

I am a Full Professor of St.Petersburg State University and a supervisor of Actuarial Training Program at the university.

I took part in the previous International Congress of Actuaries in Birmingham in 1998. My report was published in the transactions: G.V. Tchernova The Contents Nature of Underwriting in Life Insurance. Trans. 26th International Congress of Actuaries, Birmingham, UK, 7-12 June 1998, Volume 3, ISBN for 3-rd Volume 0901066 494, Printed by Bell@Bain Limited, Glasgow, UK, p.131-146. My participating was granted by the Organizational Committee of 26th ICA.