Portfolio selection with higher-order moments: A target-distribution approach
(2019) 9th General AMaMeF Conference — Location: Paris (11.June.2019)
Files
PortfolioselectionwithhigherordermomentsAtarget-distributionapproach.pdf
Closed Access - Adobe PDF
- 545.22 KB
Details
- Authors
- Abstract
- Accounting for higher moments is an important challenge in portfolio selection. To that aim, we propose to capture the investor's preferences via a target-return distribution and, in turn, consider the minimum-divergence portfolio, which is the portfolio whose return density minimizes the Kullback-Leibler divergence with respect to the target-return density. We focus on the generalized-normal target-return distribution, an extension of the Gaussian distribution allowing for non-zero excess kurtosis, and match the target-return mean and variance with those of a reference portfolio on the efficient frontier. As a result, the minimum-divergence portfolio is a shrinkage between the reference portfolio and a portfolio whose higher return moments correspond to those of the generalized-normal target-return distribution. We also recover Markowitz's efficient frontier when asset returns are Gaussian or when the target return follows a Dirac delta distribution. With regards to finite-sample estimation, we provide a closed-form estimator of the objective function based on a Gaussian-mixture density estimator. We test our minimum-divergence portfolio strategy based on several mean-variance reference portfolios from the literature and show that our portfolios provide a similar mean-variance trade-off but substantially less tail risk, including in crisis periods. Our strategy outperforms common higher-moment portfolio strategies as well.
- Affiliations
Citations
Lassance, N., & Vrins, F. (2019). Portfolio selection with higher-order moments: A target-distribution approach. 9th General AMaMeF Conference, Paris. https://hdl.handle.net/2078.5/63916