CFE 2019: Start Registration
View Submission - CMStatistics
B1497
Title: Nonlinear loss aversion and portfolio optimization Authors:  Anna Maria Fiori - University of Milano-Bicocca (Italy) [presenting]
Alessandro Avellone - University of Milano-Bicocca (Italy)
Ilaria Foroni - University of Milano-Bicocca (Italy)
Abstract: The interest for loss averse preferences in portfolio optimization has been recently rekindled by increased financial market instability and by failures of various pension funds, also reported in the mainstream media. A number of papers have shown that loss averse portfolio theory can be made consistent with decision-theoretic models of choice under uncertainty and deliver superior performance relative to traditional Mean-Variance analysis. With a few notable exceptions, the majority of these papers have focused on linear or power loss aversion (LA), solving the optimal asset allocation problem either by linear programming or by Monte Carlo simulation for very small portfolios. The proposal puts forth a methodology that can be used to derive optimal asset allocation rules for general forms of nonlinear LA. The portfolio problem is solved by a new algorithm based on Particle Swarm Optimization (PSO), which leads to computationally efficient solutions even when a high number of assets is considered. Our variant of PSO permits the inclusion of real-world constraints (e.g., buy-in thresholds, cardinality) through a repair strategy that identifies unfeasible solutions and brings them back into the feasible domain of the investment problem. An empirical study is conducted to assess the impact of behavioral parameters and to evaluate the effective ability of LA portfolios to increase investor protection in adverse market conditions.