Rare event simulation for processes generated via stochastic fixed point equations (Q744388)

A simulation algorithm of estimating the tail probability of a random variable \(V\), satisfying an equation \(V = f(V)\) in the sense of distribution, is considered, where \(f(v)=A\max(v,D)+B\), \(A,B,D\) are random variables. The stochastic fixed point problem in the sense of law typically comes from, for instance, the financial time series, the ruin problem with stochastic investments, ARCH(1) and so on. A dynamic importance sampling estimator is given, which is shown to be consistent, strongly efficient and optimal. The running time of the algorithm is described. Numerical examples are discussed.