Extreme-value based estimation of the conditional tail moment with application to reinsurance rating (Q2682980)

Each random variable \(X\) can be characterized be distribution function \(F(x)=\mathbb{P}(X\leqslant x)\), \(x\in\mathbb{R}\), and by the quantile function \[ U(x)=\inf\{y:F(y)\geqslant 1-1/x\}, x>1. \] The conditional tail moment of a non-negative random variable \(X\) is \[ \theta_{\{\beta,\,p\}}=\mathbb{E}\big(X^\beta\,|\, X>U(1/p)\big), \] provided if \(\mathbb{E}X^\beta<\infty\), where \(p\in(0,1)\) and \(\beta>0\). Let \(\{X_1,X_2,\ldots,X_n\}\) be a sample of independent and identically distributed random variables with order statistics \(\{X_{1,n}\leqslant X_{2,n}\leqslant\ldots X_{n,n}\}\) and the estimator \[ \widetilde{\theta}_{\{\beta,n\}}=\frac{1}{k}\sum_{j=1}^k X^\beta_{n-j+1,n}. \] Under certain conditions, authors of the paper derive several statements on the limiting behavior of the ratio \(\widetilde{\theta}_{\{\beta,n\}}/ \theta_{\{\beta,\,p\}}\) in case \(k,n\rightarrow\infty\) and \(k/n\rightarrow 0\). The developed methodology is applied to estimation of the expected payment and the variance of the payment under an excess-of-loss reinsurance contract.