Priors from DSGE Models for Dynamic Factor Analysis

Bäurle, Gregor (August 2008). Priors from DSGE Models for Dynamic Factor Analysis (Discussion Papers 08-03). Bern: Department of Economics

[img]
Preview
Text
dp0803.pdf - Published Version
Available under License Creative Commons: Attribution (CC-BY).

Download (423kB) | Preview

We propose a method to incorporate information from Dynamic Stochastic General Equilibrium (DSGE) models into Dynamic Factor Analysis. The method combines a rocedure previously applied for Bayesian Vector Autoregressions and a Gibbs Sampling approach for Dynamic Factor Models. The factors in the model are rotated such that they can be interpreted as variables from a DSGE model. In contrast to standard Dynamic Factor Analysis, a direct economic interpretation of the factors is given. We evaluate the forecast performance of the model with respect to the amount of information from the DSGE model included in the estimation. We conclude that using prior information from a standard New Keynesian DSGE model improves the forecast performance. We also analyze the impact of identified monetary shocks on both the factors and selected series. The interpretation of the factors as variables from the DSGE model allows us to use an identification scheme which is directly linked to the DSGE model. The responses of the factors in our application resemble
responses found using VARs. However, there are deviations from standard results when looking at the responses of specific series to common shocks.

Item Type:

Working Paper

Division/Institute:

03 Faculty of Business, Economics and Social Sciences > Department of Economics

UniBE Contributor:

Bäurle, Gregor Stefan

Subjects:

300 Social sciences, sociology & anthropology > 330 Economics

Series:

Discussion Papers

Publisher:

Department of Economics

Language:

English

Submitter:

Lars Tschannen

Date Deposited:

06 Oct 2020 16:28

Last Modified:

06 Oct 2020 16:28

JEL Classification:

C11, C15, C22, C53, E37, E47

BORIS DOI:

10.7892/boris.145711

URI:

https://boris.unibe.ch/id/eprint/145711

Actions (login required)

Edit item Edit item
Provide Feedback