Modelling dependence uncertainty in the extremes of Markov chain (Q2488432): Difference between revisions

A stationary Markov process \(x_1,\dots,x_n\) is considered. The marginal distribution of \(x_i\) above a high threshold is a generalized extreme value distribution and the joint survival function of \((x_j,x_{j+1})\) is \[ S(x_1,x_2)=[S(x_1)^{1-k}+S(x_2)^{1-k}-1]^{q/(1-k)} [S(x_1)S(x_2)]^{1-q}, \] where \(S\) is the marginal survival function, \(q\) and \(k\) are parameters. Bayesian approach is developed for the estimation of parameters and the measure of extremal dependence of \(x_i\) and \(x_{i+1}\). Results of simulation and application to rainfall data in Venezuela are presented.