Finance-Insurance
Time: 11.00 am
Date: 03th of April 2025
Room 3001

Mattheo BARIGOZZI (Università di Bologna) “Score-Driven High-Dimensional Approximate Dynamic Factor Models: Estimation and Inference”

Abstract : We propose a dynamic factor model for high-dimensional time series where the dynamics of the latent factors is non-linear and generated by a multivariate score-driven model, thus allowing to model non-linearities and heavy tails. Estimation is in two-steps: first the factors are extracted either via Principal Components or via Diversified Projections and then the parameters of the score-driven model for the estimated factors are estimated via Maximum Likelihood. Models for the conditional mean and the conditional variance are considered. Consistency and asymptotic normality for the parameters hold as both the number of time series and the sample size diverge to infinity. Moreover, valid asymptotic prediction intervals are built for the latent factors. Numerical results confirm the goodness of our estimator. 

Joint work : Enzo D’Innocenzo

Organizers:  Zakoian Jean-Michel