A general multivariate new better than used (MNBU) distribution and its properties (Q2227194)

The authors consider a multivariate shock model for a system made up of \(n\) components. Shocks arrive at the points of a generalised Pólya process, and can either affect a single system component or all \(n\) components simultaneously. Components affected by a shock immediately fail. The random variables of interest here are the time \(X_i\) (\(i=1,\ldots,n\)) until the failure of the \(i\)th component, and \(X=\min_iX_i\). Based on an explicit expressions they derive for the joint survival function of the \(X_i\), the authors investigate positive dependence properties of these random variables. For example, the \(X_i\) are shown to satisfy the positive upper orthant dependency property, and under certain conditions (that the underlying intensity function of the shocks is increasing in time and that the probability a shock affects a given component is constant in time) the \(X_i\) are multivariate new better than used. The authors conclude the paper with a discussion of the consequences of these results for the time \(X\) until failure of the system.