Hello, May I know if there is any example on how the nested stochastic model can be used? Under a stochastic model, the inputs/outputs are in the form of distribution. So the input parameters could be in the form of a distribution. How does a nested stochastic model work? Distribution in a distribution?
Hello Alex! An example of a nested stochastic model for the mobile phone insurer in the question could be a 'frequency-severity' model: Model the number of claims per annum, N, as a stochastic variable, eg with a Poisson(lambda) distribution. Within each simulation of the number of claims, model the claim amounts: X1, ..., XN as stochastic random variables, eg with logN(mu, sigma^2) distributions. This is like the compound distribution model from CT3/CT6 for modelling the total claim amount, S: S = X1+ ... + XN. Keep up the good work! Anna