Home // DATA ANALYTICS 2013, The Second International Conference on Data Analytics // View article

Efficient Optimization of Reinsurance Contracts using Discretized PBIL

Authors:
Omar Andres Carmona Cortes
Andrew Rau-Chaplin
Duane Wilson
Ian Cook
Jurgen Gaiser-Porter

Keywords: Financial Risk Management; Reinsurance Contract Optimization; Population Based Incremental Learning; Insurance and Reinsurance Analytics.

Abstract:
Risk hedging strategies are at the heart of financial risk management. As with many financial institutions, insurance companies try to hedge their risk against potentially large losses, such as those associated with natural catastrophes. Much of this hedging is facilitated by engaging in risk transfer contracts with the global reinsurance market. Devising an effective hedging strategy depends on careful data analysis and optimization. In this paper, we study from the perspective of an insurance company a Reinsurance Contract Optimization problem in which we are given a reinsurance contract consisting of a fixed number of contractual layers and a simulated set of expected loss distributions (one per layer), plus a model of reinsurance market costs. Our task is to identify optimal combinations of placements such that for a given expected return the associated risk value is minimized. The solution to this high-dimensional multi-objective data analysis and optimization problem is a Pareto frontier that quantifies the best available trade-offs between expected risk and returns. Our approach to this reinsurance contract optimization problem is to adapt the evolutionary heuristic search method called Population Based Incremental Learning, or PBIL, to work with discretized solution spaces. Our multi-threaded Discretized PBIL method (or DiPBIL) is able to solve larger “real world” problem instances than previous methods. For example, problems with a 5% discretization and 7 or less contractual layers can be solved in less than 1h:20m, while previously infeasible problems that would have taken weeks or even months to run with as many as 15 layers can be solved in less than a day.

Pages: 18 to 24

Copyright: Copyright (c) IARIA, 2013

Publication date: September 29, 2013

Published in: conference

ISSN: 2308-4464

ISBN: 978-1-61208-295-0

Location: Porto, Portugal

Dates: from September 29, 2013 to October 3, 2013