Lorenz and Excess Wealth Orders, with Applications in Reinsurance Theory (Q4512140)

The interest of the actuarial community in stochastic orderings originated in the seminal papers by \textit{K. Borch} [ASTIN Bulletin 1, 245-255 (1961)]; \textit{H. Bühlmann, B. Gagliardi, H. Gerber} and \textit{E. Straub} [ASTIN Bulletin 9, 75-83 (1977)]\ and \textit{M. J. Goovaerts, F. De Vilder} and \textit{J. Haezendonck} [Insur. Math. Econ., 1, 131-163 (1982; Zbl 0492.62090)]. These techniques have been developed in an important way and constitute now a classical tool for comparing the riskiness of different random situations. The purpose of this paper is to study the stochastic ordering by means of pointwise comparison of the Lorenz curves, or of the excess wealth transforms, of two risks. The authors start with the presentation of the Lorenz order in an actuarial framework. The main advantage of the Lorenz order is that it allows the actuary to take into account the amounts of the premium charged when two risks are compared. The Lorenz order is applied to reinsurance theory in order to derive the worst treaty for the ceding company, as well as for the reinsurer. Further, the authors study the presentation of the excess wealth order in an actuarial framework. It is shown that the excess wealth order is actually based on the comparison of the ratios of the stop-loss premiums to the direct premiums. Also, a new ordering of risks based on the preferences shared by all the ``rational'' reinsurers is then proposed. Several sufficient conditions for this ordering are provided and, finally, the results are applied to exponentially distributed risks.