Abstract
This paper presents the R package GAS for the analysis of time series under the generalized autoregressive score (GAS) framework of Creal, Koopman, and Lucas (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 non-linear models. The GAS package provides functions to simulate univariate and multivariate GAS processes, to 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 financial asset returns.
Highlights
Time-variation in the parameters describing a stochastic time series process is pervasive in almost all applied scientific fields
When we model the scale parameter of a Student-t distribution, we need to ensure its positiveness. Even if this problem can be solved by imposing constraints on ξ, the standard solution under the generalized autoregressive score (GAS) framework is to use a link function Λ(·) that maps θt ∈ J into θt and where θt ∈ J has the linear dynamic specification of (2)
The coefficients to be estimated are gathered into ξ ≡ (κ, A, B) and estimated by numerically maximizing thelikelihood function as detailed in Section 2.3.5 In Appendix B we discuss the choice of mapping functions for GAS models in more details
Summary
Time-variation in the parameters describing a stochastic time series process is pervasive in almost all applied scientific fields. Creal et al (2013) and Harvey (2013) propose to use the score of the conditional density function as the main driver of time-variation in the param-. The R (R Core Team 2018) package GAS (Catania, Boudt, and Ardia 2019) is conceived to be of relevance for the modeling of all types of time series data. It does not matter whether they are real-valued, integer-valued, (0, 1)-bounded or strictly positive, as long as there is a conditional density for which the score function and the Hessian are well-defined. Under the assumption of a Gaussian distribution, the 10% return is a strong signal of an increase in volatility, while under a fat-tailed Student-t distribution, the signal is weakened because of the higher probability that the extreme value is an observation from the tails
Sign-in/Register to view highlights & summary Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
