Volume 10 Issue 1 - March 2018

  • 1. Investigation of asset allocation performance using shrinkage estimators for higher moments

    Authors : Yesuk Jung, Ian Sutherland, Gunhee Lee

    Pages : 9-18

    DOI : http://dx.doi.org/10.21172/1.101.03

    Keywords : Shrinkage EstimatorPostmodern Portfolio TheoryHigher Order MomentsAsset AllocationPortfolio Optimization

    Abstract :

    Post-Modern Portfolio Theory extends the notion that investors’ satisfaction is reliant on minimizing downside deviation by incorporating higher order co-moments for optimizing the asset weights in a portfolio to accommodate for the difference in upside and downside deviation. Utilizing the third and fourth order co-moments of asset returns, co-skewness and co-kurtosis respectively, portfolio optimization algorithms are able to provide weights in respect to not only maximizing average return and minimizing risk in terms of covariance, but to maximize co-skewness in order to put higher weights on sets of assets whose returns tend to skew positively while minimizing co-kurtosis in order to curtail outliers. However, the use of higher order moments results in a problem of high dimensionality, where there are many parameters to estimate providing unstable estimators with large standard error. This study assesses the performance of several shrinkage estimators in order to overcome the dimensionality problems and to investigate the performances of several feasible optimization models using S&P 500 data. Our study shows that the optimization method with shrinkage estimators for higher order co-moments is superior to the traditional Markowitz’s mean-variance optimization. It can also be concluded that applying shrinkage estimators and higher order co-moments to the problem of portfolio optimization can play an important role in improving asset allocations.

    Citing this Journal Article :

    Yesuk Jung, Ian Sutherland, Gunhee Lee, "Investigation of asset allocation performance using shrinkage estimators for higher moments", https://www.ijltet.org/journal_details.php?id=928&j_id=4371, Volume 10 Issue 1 - March 2018, 9-18, #ijltetorg