Optimal reinsurance with positively dependent risks (Q2427807)

Assume that the insurance portfolio has \(n\) policy holders. The loss for the \(i\)-th (\(i=1,2,\dots, n\)) policy holder is a random variable \(X_i\). To protect it from a potential huge loss, the insurer applies a reinsurance strategy \(I_i\) to the loss of policy \(i\), where the function \(I_i(x),x\geq 0,\) is increasing and satisfies \(0\leq I_i(x)\leq x\) for each \(i=1,2,\dots, n\). The total retained loss of the insurer is \(\;I_1(X_1)+I_2(X_2)+\dots +I_n(X_n)\;\) for some \(n\)-dimensional individualized reinsurance contract \(\;(I_1,I_2,\dots, I_n)\;\). In the paper, it is considered a situation in which the losses \(\;X_1,X_2,\dots, X_n\;\) form positively dependent through the stochastic ordering random vector \(\;(X_1,X_2,\dots, X_n)\). It is shown that the excess-of-loss treaty is the optimal reinsurance form among a general class of individualized reinsurance contracts. The obtained results extend the study of \textit{M. Denuit} and \textit{C. Vermandele} [Insur. Math. Econ. 22, No. 3, 229--233 (1998; Zbl 0986.62085)] of individualized reinsurance treaties to dependent risks.