A numerical method for the estimation of time-varying parameter models in large dimensions

TL;DR

This paper introduces a computationally efficient numerical method for estimating large-scale time-varying parameter models, improving stability and reducing costs in high-dimensional settings with many covariates.

Contribution

It develops a novel stable orthogonal transformation-based approach for large TVP models, extending to rolling window estimation and exploiting sparse structures for efficiency.

Findings

Effective in high dimensions with many covariates

Reduces computational costs through sparse structure exploitation

Improves stability and accuracy of estimates

Abstract

A novel numerical method for the estimation of large time-varying parameter (TVP) models is proposed. The updating and smoothing estimates of the TVP model are derived within the context of generalised linear least squares and through numerically stable orthogonal transformations. The method developed is based on computationally efficient strategies. The computational cost is reduced by exploiting the special sparse structure of the TVP model and by utilising previous computations. The proposed method is also extended to the rolling window estimation of the TVP model. Experimental results show the effectiveness of the new updating, window and smoothing strategies in high dimensions when a large number of covariates and regressions are included in the TVP model.