Variational Bayes Inference in High-Dimensional Time-Varying Parameter Models

Koop, Gary; Korobilis, Dimitris

doi:10.2139/ssrn.3246472

Cited by 27 publications

(24 citation statements)

References 63 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…large, then MCMC methods may be computationally infeasible. In such a case, variational Bayesian methods may be a practical alternative (see Koop and Korobilis, 2018). But with variational Bayes methods, the SAVS algorithm would be applied on the approximate posterior mean and model uncertainty ignored.…”

Section: The Tvp Regression Modelmentioning

confidence: 99%

Inducing Sparsity and Shrinkage in Time-Varying Parameter Models

Huber

Koop

Onorante

2020

Journal of Business & Economic Statistics

Self Cite

View full text Add to dashboard Cite

Time-varying parameter (TVP) models have the potential to be over-parameterized, particularly when the number of variables in the model is large. Global-local priors are increasingly used to induce shrinkage in such models. But the estimates produced by these priors can still have appreciable uncertainty. Sparsification has the potential to remove this uncertainty and improve forecasts. In this paper, we develop computationally simple methods which both shrink and sparsify TVP models. In a simulated data exercise we show the benefits of our shrink-then-sparsify approach in a variety of sparse and dense TVP regressions. In a macroeconomic forecast exercise, we find our approach to substantially improve forecast performance relative to shrinkage alone.

show abstract

Section: The Tvp Regression Modelmentioning

confidence: 99%

Inducing Sparsity and Shrinkage in Time-Varying Parameter Models

Huber

Koop

Onorante

2020

Journal of Business & Economic Statistics

Self Cite

View full text Add to dashboard Cite

show abstract

“…We hope that our results will encourage the use of variational methods in fields like applied statistics and econometrics where variational inference remains somewhat underutilized despite its tremendous impact in machine learning. Recent applications of variational inference in econometrics include Bonhomme (2021), Mele and Zhu (2019), and Koop and Korobilis (2018).…”

Section: Related Workmentioning

confidence: 99%

On the Robustness to Misspecification of $α$-Posteriors and Their Variational Approximations

Medina,

Olea,

Rush

et al. 2021

Preprint

View full text Add to dashboard Cite

α-posteriors and their variational approximations distort standard posterior inference by downweighting the likelihood and introducing variational approximation errors. We show that such distortions, if tuned appropriately, reduce the Kullback-Leibler (KL) divergence from the true, but perhaps infeasible, posterior distribution when there is potential parametric model misspecification. To make this point, we derive a Bernstein-von Mises theorem showing convergence in total variation distance of α-posteriors and their variational approximations to limiting Gaussian distributions. We use these distributions to evaluate the KL divergence between true and reported posteriors. We show this divergence is minimized by choosing α strictly smaller than one, assuming there is a vanishingly small probability of model misspecification. The optimized value becomes smaller as the the misspecification becomes more severe. The optimized KL divergence increases logarithmically in the degree of misspecification and not linearly as with the usual posterior.

show abstract

“…If av = "dma", then the original DMA averaging scheme is performed. If av = "mse" then predictive densities in Equation 5are replaced by the inverses of Mean Squared Errors of the models [175,176]. If av = "hr1", then they are replaced by Hit Ratios (assuming time-series are in levels).…”

Section: Fundamental Functionsmentioning

confidence: 99%

Dynamic Model Averaging in Economics and Finance with fDMA: A Package for R

Drachal

2020

Signals

View full text Add to dashboard Cite

The described R package allows to estimate Dynamic Model Averaging (DMA), Dynamic Model Selection (DMS) and Median Probability Model. The original methods, and additionally, some selected modifications of these methods are implemented. For example the user can choose between recursive moment estimation and exponentially moving average for variance updating in the base DMA. Moreover, inclusion probabilities can be computed in a way using “Google Trends” data. The code is written with respect to minimise the computational burden, which is quite an obstacle for DMA algorithm if numerous variables are used. For example, this package allows for parallel computations and implementation of the Occam’s window approach. However, clarity and readability of the code, and possibility for an R-familiar user to make his or her own small modifications in reasonably small time and with low effort are also taken under consideration. Except that, some alternative (benchmark) forecasts can also be quickly performed within this package. Indeed, this package is designed in a way that is hoped to be especially useful for practitioners and researchers in economics and finance.

show abstract

Variational Bayes Inference in High-Dimensional Time-Varying Parameter Models

Cited by 27 publications

References 63 publications

Inducing Sparsity and Shrinkage in Time-Varying Parameter Models

Inducing Sparsity and Shrinkage in Time-Varying Parameter Models

On the Robustness to Misspecification of $α$-Posteriors and Their Variational Approximations

Dynamic Model Averaging in Economics and Finance with fDMA: A Package for R

Contact Info

Product

Resources

About