Hybrid multi-objective evolutionary computation of constrained downside risk-return efficient sets for credit portfolios

In contemporary credit portfolio management, the portfolio risk-return analysis of financial instruments using certain downside credit risk measures requires the computation of a set of Pareto-efficient portfolio structures in a non-linear, non-convex setting. For real-world problems, additional constraints, e. g. supervisory capital limits, have to be respected. Particularly for formerly non-traded instruments, e. g. corporate loans, a discrete set of decision alternatives has to be considered for each financial instrument. The main result of this paper is a new, fast and flexible framework for solving the above issues using a hybrid heuristic method that combines multi-objective evolutionary and problem-specific local search methods in a unique way. We explicitly incorporate computational complexity in some of our considerations and consider proper genetic modelling of portfolio credit risk related problems. Also, we analyse empirical results from a study based on our implementation of the proposed hybrid method in a specific portfolio credit risk model context. These results show that this method is superior in convergence speed to a non-hybrid evolutionary approach and that our implementation finds risk-return efficient sets within reasonable time.