Quadratic hedging: an actuarial view extended to solvency control (Q362036)

In an incomplete financial market, a popular way to hedge a portfolio is quadratic risk minimisation. If the target function is taken with respect to the physical measure, one has to use the local risk minimisation introduced by \textit{M. Schweizer} [Stochastic Processes Appl. 37, No. 2, 339--363 (1991; Zbl 0735.90028)]. This approach has the advantage that one often does not know the pricing measure, but can easily estimate the physical measure. Under the pricing measure, \textit{H. Föllmer} and \textit{D. Sondermann} [in: Contributions to mathematical economics, Hon. G. Debreu, 206--223 (1986; Zbl 0663.90006)] introduced the quadratic hedging by minimising the expected square of the costs of a hedging strategy. In this paper, the Föllmer and Sondermann case is treated. It is shown that the problem can be considered as in credibility theory as an orthogonal projection. The setup is extended to solvency control. It is assumed that the surplus has to exceed an externally given minimal solvency process. As an example, the theory is applied to a model for catastrophe bonds. Two dependent risk processes are given to model a catastrophe loss index and the losses from the portfolio of an insurer. The quadratic hedging then determines the optimal number of catastrophe bonds the insurer should have in the portfolio.