最坏损失度量方法.pdfVIP

Worst case risk measurement: back to the future?∗ In honor of Etienne De Vijlder Roger J.A. Laeven‡,† , Marc J. Goovaerts‡,§ , Rob Kaas‡ ‡ University of Amsterdam, Dept. of Quantitative Economics, Roetersstraat 11, 1018 WB Amsterdam, The Netherlands § Catholic University of Leuven, Dept. of Applied Economics, Naamsestraat 69, B-3000 Leuven, Belgium September 21, 2006 Abstract This paper studies the problem of finding best-possible upper bounds on risk mea- sures under incomplete probabilistic information. Both the case of univariate risk and the case of multivariate risk are considered. Furthermore, we aim to identify the probability distributions that give rise to the worst case scenarios. The problem of worst case risk measurement has been studied extensively by Etienne De Vijlder and his co-authors. We review and extend some of their work. To our regret, Etienne passed away in the beginning of the year 2004. As a sign of gratitude for his large contribution to Actuarial Science, we dedicate this work to him. Keywords: Risk measurement, Generalized scenarios, Worst case scenario, Cones, Linear programming JEL-Classification: D81, G10, G20 MSC-Classification: 60E05, 62P05, 90C05 ∗We thank Paul Embrechts for inviting us to write this tribute to Etienne De Vijlder. †Corresponding author. E-mail: R.J.A.Laeven@uva.nl, Phone: +31 20 525 7317, Fax: +31 20 525 4349. 1 1 Introduction This paper studies the problem of finding best-possible upper bounds on risk measures when there is incomplete probabilistic information on th