Description: When fitting a model to any data, there is some uncertainty about which model is best. Green [1995] quantifies this uncertainty through the Reversible Jump Markov Chain Monte Carlo [RJMCMC] method. When using the RJMCMC method in a regime-switching situation, the chain determines the optimal number of regimes by jumping between various possibilities. This method gives each model its posterior probability of being the best. After an overview of the methodology, the authors apply it to various datasets and discuss the applications in modern actuarial science.

Description: One of the first projects undertaken by Claim Analytics was to create a model to predict return to work for group insurance claimants with long term disabilities. This report, sponsored in part by the Society of Actuaries Health Section , is a case study on creating such a model.

Description: This paper shows how actuarial pricing based on the mathematical concept of present value can be formulated and solved using Markov chains and an approach borrowed from engineering mathematics called ballistic control theory.

Description: The actuarial profession is faced with a problem of determining delta-ized reserves on a basis which will include appropriate provision for adverse deviations in mortality, lapses, interest rates, and expense rates. This paper describes briefly a method which could be used to solve this Followed by papers submitted by 2 authors problem in a flexible manner.

Description: This paper describes risk-factor-dependent multiple decrements and actuarial cost functions as a general compartment model, a discrete time and discrete state stochastic model of morbidity and mortality processes. For large numbers of risk factors, the model is extended to its natural continuous time and state form. From Transactions of Society of Actuaries 1991, Vol. 43.

Description: This paper presents diagnostics and tests for tile retrospective identification of abrupt change in the mean level of a process where the mean and variance of the underlying random variables cannot be functionally separated. From Actuarial Research Clearing House 1994, VOL. 1.

Description: This paper endeavors to obtain the explicit formulas of the probability of ruin in a Markov-modulated model where claim intensities, claim sizes and premiums vary according to a Markovian environment.

Description: In this paper, the authors show how probabilities and actuarial values may be calculated using a time-homogeneous Markov model. Piecewise constant transition intensities are suggested as a way to extend the approach to the inhomogeneous case. From Actuarial Research Clearing House 1993, VOL. 1.