Approximate Bayesian inference and forecasting in huge-dimensional multi-country VARs

TL;DR

This paper introduces fast Bayesian methods for estimating large multi-country PVARs, leveraging Gaussian approximations and coefficient grouping to enable efficient and accurate economic forecasting.

Contribution

It develops a novel integrated rotated Gaussian approximation approach for scalable Bayesian estimation in high-dimensional PVARs, improving computational efficiency.

Findings

The method produces competitive forecasts rapidly.

Grouping coefficients enhances estimation efficiency.

The approach handles huge models of the world economy.

Abstract

Panel Vector Autoregressions (PVARs) are a popular tool for analyzing multi-country datasets. However, the number of estimated parameters can be enormous, leading to computational and statistical issues. In this paper, we develop fast Bayesian methods for estimating PVARs using integrated rotated Gaussian approximations. We exploit the fact that domestic information is often more important than international information and group the coefficients accordingly. Fast approximations are used to estimate the latter while the former are estimated with precision using Markov chain Monte Carlo techniques. We illustrate, using a huge model of the world economy, that it produces competitive forecasts quickly.