CMStatistics 2016: Start Registration
View Submission - CMStatistics
B0852
Title: Estimation of Higher Order Comoments using Unobserved Factors Authors:  Dries Cornilly - Vrije Universiteit Brussel (Belgium) [presenting]
Tim Verdonck - KU Leuven and UAntwerpen - imec (Belgium)
Kris Boudt - Vrije Universiteit Brussel and VU Amsterdam (Belgium)
Abstract: Reliable estimates of the higher order moments of the distribution of financial portfolio returns are needed for allocation decisions as well as assessing their risk. Sample based estimation methods for the higher order comoments of the underlying assets suffer from a curse of dimensionality, introducing large estimation errors and yielding unstable estimates. To counter this, we propose to estimate the higher order comoments structurally. The comoments induced by a factor model with unobserved factors are fitted to initial estimates by means of generalized least squares. In an empirical application, we show our estimator to outperform the sample estimator and the one-factor shrinkage estimator in portfolio allocation based on maximizing utility under CVaR constraints.