scientific article; zbMATH DE number 4026543

From MaRDI portal
Publication:3768161





Two general results are shown on characterizations of continuous distributions based on the form of appropriate conditional expectations. It is shown specifically that for a continuous random variable X on \([\alpha,\beta)\) the functional form of \(E[h(X)| X>z]\) uniquely determines the functional form of \(F(x)\), the distribution function of X. The function \(h(x)\) is assumed to be real-valued, monotone, continuous with a continuous derivative on \((\alpha,\beta)\) and satisfying \[ \lim_{x\to \beta^-}h(x)=\pm \infty \quad and\quad E(h(X))=h(\alpha)+c,\quad c>0. \] If the last two assumptions are modified to \[ \lim_{x\to \alpha^+}h(x)=\pm \infty \quad and\quad E(h(X))=h(\beta)+c,\quad c>0, \] respectively, then it is shown that the functional form of \(E[h(X)| x\leq z]\) uniquely determines the functional form of \(F(x)\). Appropriate forms of \(h(x)\) lead to characterizations of the Weibull, Pareto, beta, logistic and Burr distributions.











This page was built for publication:

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q3768161)