No label defined (Q4416727)

The distribution function of the generalized Burr-Gamma (GBG) distribution is \[ F(x):=(\Gamma(k))^{-1} \int_0^{v_{\xi}(x)} e^{-u} u^{k-1} du, \] where \(v_{\xi}(x):={\xi}^{-1} \log (1+\xi v(x))>0\) and either \(v(x)=(x e^{\alpha})^1/{\beta}\) or \(v(x)=\exp\{\psi(k)+(\log x+u)\sigma^{-1} \sqrt{\psi'(k)}\}\); here \(\psi(k)\) and \(\psi'(k)\) are the digamma and trigamma functions. This family includes as special cases the Gumbel, Weibull, Gompertz and Burr distributions as well as the generalized Pareto distribution. The authors study the mean and variance of the GBG distribution and derive the maximum data information prior for it. They apply it to two examples of extreme value inference: hurricanes in the USA and the water inflow in the Gariep dam in South Africa. For the entire collection see [Zbl 1009.00022].