Date: Tuesday, 11 June 2024, at 12:15 pm
Location: Seminar Room Dono Giannessi, DEM
Speaker and Title:
Emilija Dzuverovic (University of Pisa)
Identification and Dynamic Loadings in Score-driven Factor Models
Discussant: Luca Trapin, University of Bologna
Abstract:
We show that, for a certain class of scaling matrices, including the inverse square root of the conditional Fisher information, score-driven factor models are identifiable up to a multiplicative scalar constant. This gives them a clear advantage in terms of economic interpretability with respect to parameter-driven factor models, which are identifiable only up to rotations. Moreover, the identification of the factors is order-invariant.
We then show that this result can be extended to score-driven factor models with dynamic loadings. To make the estimation of such models feasible in large dimensions, we propose an iterative two-step estimation algorithm that simplifies considerably the inference of the parameters. We test our results and the iterative two-step approach using simulated and empirical data. Joint work with Fulvio Corsi (University of Pisa) and Giuseppe Buccheri (University of Verona).
Download area