Generalized Autoregressive Score Models in R: The GAS Package

TL;DR

The paper introduces the GAS R package for modeling and forecasting time series using the Generalized Autoregressive Score framework, enabling simulation, estimation, and forecasting of univariate and multivariate processes with applications to financial data.

Contribution

It provides the first comprehensive R implementation of the GAS framework, including functions for simulation, estimation, and forecasting of complex time series models.

Findings

Demonstrates the package's effectiveness through a financial assets case study.

Shows accurate estimation of time-varying conditional densities.

Facilitates flexible modeling of nonlinear time series processes.

Abstract

This paper presents the R package GAS for the analysis of time series under the Generalized Autoregressive Score (GAS) framework of Creal et al. (2013) and Harvey (2013). The distinctive feature of the GAS approach is the use of the score function as the driver of time-variation in the parameters of nonlinear models. The GAS package provides functions to simulate univariate and multivariate GAS processes, estimate the GAS parameters and to make time series forecasts. We illustrate the use of the GAS package with a detailed case study on estimating the time-varying conditional densities of a set of financial assets.