Computationally efficient inference in large Bayesian mixed frequency VARs

Deborah Gefang; Gary Koop; Aubrey Poon

Computationally efficient inference in large Bayesian mixed frequency VARs

Research output: Contribution to journal › Article › peer-review

Abstract

Mixed frequency Vector Autoregressions (MF-VARs) can be used to provide timely and high frequency estimates or nowcasts of variables for which data is available at a low frequency. Bayesian methods are commonly used with MF-VARs to overcome over-parameterization concerns. But Bayesian methods typically rely on computationally demanding Markov Chain Monte Carlo (MCMC) methods. In this paper, we develop Variational Bayes (VB) methods for use with MF-VARs using Dirichlet-Laplace global-local shrinkage priors. We show that these methods are accurate and computationally much more efficient than MCMC in two empirical applications involving large MF-VARs.

Original language	English
Journal	Economics Letters
Publication status	Accepted/In press - 25 Mar 2020

Keywords

mixed frequency
variational inference (VI)
vector autoregression
stochastic volatility
hierarchical prior
forecasting

Access to Document

https://www.journals.elsevier.com/economics-letters

Embargoed Document

Gefang-etal-EL-2020-Computationally-efficient-inference-in-large-Bayesian-mixed-frequency-VARs
Accepted author manuscript, 383 KB
Licence: CC BY-NC-ND 4.0
Embargo ends: 25/09/21

Fingerprint Dive into the research topics of 'Computationally efficient inference in large Bayesian mixed frequency VARs'. Together they form a unique fingerprint.

Gary Koop
- gary.koop@strath.ac.uk
- Economics - Professor
- SIRE
Person: Academic

Cite this

@article{b5713ca0f81f4413817bb5147ae2c33b,

title = "Computationally efficient inference in large Bayesian mixed frequency VARs",

abstract = "Mixed frequency Vector Autoregressions (MF-VARs) can be used to provide timely and high frequency estimates or nowcasts of variables for which data is available at a low frequency. Bayesian methods are commonly used with MF-VARs to overcome over-parameterization concerns. But Bayesian methods typically rely on computationally demanding Markov Chain Monte Carlo (MCMC) methods. In this paper, we develop Variational Bayes (VB) methods for use with MF-VARs using Dirichlet-Laplace global-local shrinkage priors. We show that these methods are accurate and computationally much more efficient than MCMC in two empirical applications involving large MF-VARs.",

keywords = "mixed frequency, variational inference (VI), vector autoregression, stochastic volatility, hierarchical prior, forecasting",

author = "Deborah Gefang and Gary Koop and Aubrey Poon",

year = "2020",

month = mar,

day = "25",

language = "English",

journal = "Economic Letters",

issn = "0165-1765",

}

TY - JOUR

T1 - Computationally efficient inference in large Bayesian mixed frequency VARs

AU - Gefang, Deborah

AU - Koop, Gary

AU - Poon, Aubrey

PY - 2020/3/25

Y1 - 2020/3/25

N2 - Mixed frequency Vector Autoregressions (MF-VARs) can be used to provide timely and high frequency estimates or nowcasts of variables for which data is available at a low frequency. Bayesian methods are commonly used with MF-VARs to overcome over-parameterization concerns. But Bayesian methods typically rely on computationally demanding Markov Chain Monte Carlo (MCMC) methods. In this paper, we develop Variational Bayes (VB) methods for use with MF-VARs using Dirichlet-Laplace global-local shrinkage priors. We show that these methods are accurate and computationally much more efficient than MCMC in two empirical applications involving large MF-VARs.

AB - Mixed frequency Vector Autoregressions (MF-VARs) can be used to provide timely and high frequency estimates or nowcasts of variables for which data is available at a low frequency. Bayesian methods are commonly used with MF-VARs to overcome over-parameterization concerns. But Bayesian methods typically rely on computationally demanding Markov Chain Monte Carlo (MCMC) methods. In this paper, we develop Variational Bayes (VB) methods for use with MF-VARs using Dirichlet-Laplace global-local shrinkage priors. We show that these methods are accurate and computationally much more efficient than MCMC in two empirical applications involving large MF-VARs.

KW - mixed frequency

KW - variational inference (VI)

KW - vector autoregression

KW - stochastic volatility

KW - hierarchical prior

KW - forecasting

UR - https://www.journals.elsevier.com/economics-letters

M3 - Article

JO - Economic Letters

JF - Economic Letters

SN - 0165-1765

ER -