Choosing Prior Hyperparameters: With Applications to Time-Varying Parameter Models

Pooyan Amir-Ahmadi, Christian Matthes, Mu Chun Wang

Research output: Contribution to journalArticle

Abstract

Time-varying parameter models with stochastic volatility are widely used to study macroeconomic and financial data. These models are almost exclusively estimated using Bayesian methods. A common practice is to focus on prior distributions that themselves depend on relatively few hyperparameters such as the scaling factor for the prior covariance matrix of the residuals governing time variation in the parameters. The choice of these hyperparameters is crucial because their influence is sizeable for standard sample sizes. In this article, we treat the hyperparameters as part of a hierarchical model and propose a fast, tractable, easy-to-implement, and fully Bayesian approach to estimate those hyperparameters jointly with all other parameters in the model. We show via Monte Carlo simulations that, in this class of models, our approach can drastically improve on using fixed hyperparameters previously proposed in the literature. Supplementary materials for this article are available online.

Original languageEnglish (US)
Pages (from-to)124-136
Number of pages13
JournalJournal of Business and Economic Statistics
Volume38
Issue number1
DOIs
StatePublished - Jan 2 2020

Fingerprint

Time-varying Parameters
Hyperparameters
Stochastic Volatility
Model
Financial Data
Scaling Factor
Macroeconomics
Bayesian Methods
Hierarchical Model
Prior distribution
Bayesian Approach
scaling
macroeconomics
Covariance matrix
Sample Size
Monte Carlo Simulation
time
Time-varying parameter model
simulation
Estimate

Keywords

  • Bayesian VAR
  • Bayesian inference
  • Time variation

ASJC Scopus subject areas

  • Statistics and Probability
  • Social Sciences (miscellaneous)
  • Economics and Econometrics
  • Statistics, Probability and Uncertainty

Cite this

Choosing Prior Hyperparameters : With Applications to Time-Varying Parameter Models. / Amir-Ahmadi, Pooyan; Matthes, Christian; Wang, Mu Chun.

In: Journal of Business and Economic Statistics, Vol. 38, No. 1, 02.01.2020, p. 124-136.

Research output: Contribution to journalArticle

@article{0860cc23242746a9941045960e32bcdb,
title = "Choosing Prior Hyperparameters: With Applications to Time-Varying Parameter Models",
abstract = "Time-varying parameter models with stochastic volatility are widely used to study macroeconomic and financial data. These models are almost exclusively estimated using Bayesian methods. A common practice is to focus on prior distributions that themselves depend on relatively few hyperparameters such as the scaling factor for the prior covariance matrix of the residuals governing time variation in the parameters. The choice of these hyperparameters is crucial because their influence is sizeable for standard sample sizes. In this article, we treat the hyperparameters as part of a hierarchical model and propose a fast, tractable, easy-to-implement, and fully Bayesian approach to estimate those hyperparameters jointly with all other parameters in the model. We show via Monte Carlo simulations that, in this class of models, our approach can drastically improve on using fixed hyperparameters previously proposed in the literature. Supplementary materials for this article are available online.",
keywords = "Bayesian VAR, Bayesian inference, Time variation",
author = "Pooyan Amir-Ahmadi and Christian Matthes and Wang, {Mu Chun}",
year = "2020",
month = "1",
day = "2",
doi = "10.1080/07350015.2018.1459302",
language = "English (US)",
volume = "38",
pages = "124--136",
journal = "Journal of Business and Economic Statistics",
issn = "0735-0015",
publisher = "Taylor & Francis",
number = "1",

}

TY - JOUR

T1 - Choosing Prior Hyperparameters

T2 - With Applications to Time-Varying Parameter Models

AU - Amir-Ahmadi, Pooyan

AU - Matthes, Christian

AU - Wang, Mu Chun

PY - 2020/1/2

Y1 - 2020/1/2

N2 - Time-varying parameter models with stochastic volatility are widely used to study macroeconomic and financial data. These models are almost exclusively estimated using Bayesian methods. A common practice is to focus on prior distributions that themselves depend on relatively few hyperparameters such as the scaling factor for the prior covariance matrix of the residuals governing time variation in the parameters. The choice of these hyperparameters is crucial because their influence is sizeable for standard sample sizes. In this article, we treat the hyperparameters as part of a hierarchical model and propose a fast, tractable, easy-to-implement, and fully Bayesian approach to estimate those hyperparameters jointly with all other parameters in the model. We show via Monte Carlo simulations that, in this class of models, our approach can drastically improve on using fixed hyperparameters previously proposed in the literature. Supplementary materials for this article are available online.

AB - Time-varying parameter models with stochastic volatility are widely used to study macroeconomic and financial data. These models are almost exclusively estimated using Bayesian methods. A common practice is to focus on prior distributions that themselves depend on relatively few hyperparameters such as the scaling factor for the prior covariance matrix of the residuals governing time variation in the parameters. The choice of these hyperparameters is crucial because their influence is sizeable for standard sample sizes. In this article, we treat the hyperparameters as part of a hierarchical model and propose a fast, tractable, easy-to-implement, and fully Bayesian approach to estimate those hyperparameters jointly with all other parameters in the model. We show via Monte Carlo simulations that, in this class of models, our approach can drastically improve on using fixed hyperparameters previously proposed in the literature. Supplementary materials for this article are available online.

KW - Bayesian VAR

KW - Bayesian inference

KW - Time variation

UR - http://www.scopus.com/inward/record.url?scp=85048809976&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85048809976&partnerID=8YFLogxK

U2 - 10.1080/07350015.2018.1459302

DO - 10.1080/07350015.2018.1459302

M3 - Article

AN - SCOPUS:85048809976

VL - 38

SP - 124

EP - 136

JO - Journal of Business and Economic Statistics

JF - Journal of Business and Economic Statistics

SN - 0735-0015

IS - 1

ER -