Time varying dimension models

Gary Koop, Rodney Strachan, Roberto Leon-Gonzalez, Joshua Chan

Research output: Contribution to journalArticle

Abstract

Time varying parameter (TVP) models have enjoyed an increasing popularity in empirical macroeconomics. However, TVP models are parameter-rich and risk over-fitting unless the dimension of the model is small. Motivated by this worry, this paper proposes several Time Varying dimension (TVD) models where the dimension of the model can change over time, allowing for the model to automatically choose a more parsimonious TVP representation, or to switch between different parsimonious representations. Our TVD models all fall in the category of dynamic mixture models. We discuss the properties of these models and present methods for Bayesian inference. An application involving US inflation forecasting illustrates and compares the different TVD models. We find our TVD approaches exhibit better forecasting performance than several standard benchmarks and shrink towards parsimonious specifications.
LanguageEnglish
JournalJournal of Business and Economic Statistics
Publication statusAccepted/In press - 2012

Fingerprint

Time-varying
Time-varying Parameters
Model
Forecasting
time
Overfitting
Macroeconomics
Bayesian inference
Mixture Model
Inflation
inflation
Switch
Dynamic Model
macroeconomics
popularity
Choose
Benchmark
Specification
performance

Keywords

  • Bayesian models
  • time varying parameter model

Cite this

Koop, G., Strachan, R., Leon-Gonzalez, R., & Chan, J. (Accepted/In press). Time varying dimension models. Journal of Business and Economic Statistics.
Koop, Gary ; Strachan, Rodney ; Leon-Gonzalez, Roberto ; Chan, Joshua. / Time varying dimension models. In: Journal of Business and Economic Statistics. 2012.
@article{a3631de58b98417782472c41e64d73c3,
title = "Time varying dimension models",
abstract = "Time varying parameter (TVP) models have enjoyed an increasing popularity in empirical macroeconomics. However, TVP models are parameter-rich and risk over-fitting unless the dimension of the model is small. Motivated by this worry, this paper proposes several Time Varying dimension (TVD) models where the dimension of the model can change over time, allowing for the model to automatically choose a more parsimonious TVP representation, or to switch between different parsimonious representations. Our TVD models all fall in the category of dynamic mixture models. We discuss the properties of these models and present methods for Bayesian inference. An application involving US inflation forecasting illustrates and compares the different TVD models. We find our TVD approaches exhibit better forecasting performance than several standard benchmarks and shrink towards parsimonious specifications.",
keywords = "Bayesian models, time varying parameter model",
author = "Gary Koop and Rodney Strachan and Roberto Leon-Gonzalez and Joshua Chan",
year = "2012",
language = "English",
journal = "Journal of Business and Economic Statistics",
issn = "0735-0015",

}

Time varying dimension models. / Koop, Gary; Strachan, Rodney; Leon-Gonzalez, Roberto; Chan, Joshua.

In: Journal of Business and Economic Statistics, 2012.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Time varying dimension models

AU - Koop, Gary

AU - Strachan, Rodney

AU - Leon-Gonzalez, Roberto

AU - Chan, Joshua

PY - 2012

Y1 - 2012

N2 - Time varying parameter (TVP) models have enjoyed an increasing popularity in empirical macroeconomics. However, TVP models are parameter-rich and risk over-fitting unless the dimension of the model is small. Motivated by this worry, this paper proposes several Time Varying dimension (TVD) models where the dimension of the model can change over time, allowing for the model to automatically choose a more parsimonious TVP representation, or to switch between different parsimonious representations. Our TVD models all fall in the category of dynamic mixture models. We discuss the properties of these models and present methods for Bayesian inference. An application involving US inflation forecasting illustrates and compares the different TVD models. We find our TVD approaches exhibit better forecasting performance than several standard benchmarks and shrink towards parsimonious specifications.

AB - Time varying parameter (TVP) models have enjoyed an increasing popularity in empirical macroeconomics. However, TVP models are parameter-rich and risk over-fitting unless the dimension of the model is small. Motivated by this worry, this paper proposes several Time Varying dimension (TVD) models where the dimension of the model can change over time, allowing for the model to automatically choose a more parsimonious TVP representation, or to switch between different parsimonious representations. Our TVD models all fall in the category of dynamic mixture models. We discuss the properties of these models and present methods for Bayesian inference. An application involving US inflation forecasting illustrates and compares the different TVD models. We find our TVD approaches exhibit better forecasting performance than several standard benchmarks and shrink towards parsimonious specifications.

KW - Bayesian models

KW - time varying parameter model

UR - http://www.rcfea.org/RePEc/pdf/wp44_10.pdf

UR - http://www.strath.ac.uk/media/departments/economics/researchdiscussionpapers/2011/11-16_Final.pdf

M3 - Article

JO - Journal of Business and Economic Statistics

T2 - Journal of Business and Economic Statistics

JF - Journal of Business and Economic Statistics

SN - 0735-0015

ER -