Template-Type: ReDIF-Paper 1.0
Series: Tinbergen Institute Discussion Papers
Creation-Date: 2021-01-24
Revision-Date: 2023-07-11
Number: 21-010/III
Author-Name: Andre Lucas
Author-Workplace-Name: Vrije Universiteit Amsterdam
Author-Name: Anne Opschoor
Author-Workplace-Name: Vrije Universiteit Amsterdam
Author-Name: Luca Rossini
Author-Workplace-Name: University of Milan
Title: Tail Heterogeneity for Dynamic Covariance Matrices:
the F-Riesz Distribution
Abstract: We introduce a novel model for the dynamics of fat-tailed (realized) covariance-matrix-valued time series using the new F-Riesz distribution. The model allows for
different tail behavior across the coordinates of the covariance matrix via two vector-valued degrees of freedom parameters, thus generalizing the familiar Wishart and
matrix-F distributions by introducing heterogeneous tail behavior. We show that the
filter implied by the new model is invertible and that a two-step targeted maximum
likelihood estimator is consistent. Applying the new F-Riesz model to U.S. stocks,
both tail-heterogeneity and tail-fatness are empirically relevant and produce large in-sample and out-of-sample likelihood increases and lower ex-post portfolio standard
deviations compared to static models or models with homogeneous tail behavior
Classification-JEL: C32, C58, G17
Keywords: matrix distributions, tail heterogeneity, (inverse) Riesz, fat-tails, realized
covariance matrices
File-URL: https://papers.tinbergen.nl/21010.pdf
File-Format: application/pdf
File-Size: 789.998 bytes
Handle: RePEc:tin:wpaper:20210010