From Calibration to Estimation: A Modern Bayesian Framework for Quarterly Projection Models
Andrés González Gómez and
Juan David Prada Sarmiento
No 2026/241, IMF Working Papers from International Monetary Fund
Abstract:
Quarterly Projection Models (QPMs) are widely used at central banks but rarely estimated by full Bayesian methods, because rational-expectations solvers and shock-variance priors resist gradient-based inference. We develop a workf low pairing the structure-preserving doubling algorithm, automatic dif ferentiation and the No-U-Turn Sampler with a signal-to-noise-ratio prior calibration refined by a Koopman disturbance-smoother EM step. Applied to an open-economy QPM for Brazil (2004Q1–2025Q2), it delivers stable posterior inference, tightens monetary-policy impulse-response bands relative to the prior, places the output gap inside the Banco Central do Brasil’s published envelope in about 84% of the sample, and produces stable pseudo-real-time forecasts, with its clearest edge over a Bayesian VAR on the policy rate.
Keywords: Bayesian estimation; Hamiltonian Monte Carlo; quarterly projection model; semi-structural models; monetary policy; Brazil; sacrifice ratio; natural interest rate (search for similar items in EconPapers)
Pages: 77
Date: 2026-11-06
References: Add references at CitEc
Citations:
Downloads: (external link)
https://www.imf.org/en/publications/longres?sk=580820 (application/pdf)
Our link check indicates that this URL is bad, the error code is: 404 Not Found
Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.
Export reference: BibTeX
RIS (EndNote, ProCite, RefMan)
HTML/Text
Persistent link: https://EconPapers.repec.org/RePEc:imf:imfwpa:2026/241
Ordering information: This working paper can be ordered from
http://www.imf.org/external/pubs/pubs/ord_info.htm
Access Statistics for this paper
More papers in IMF Working Papers from International Monetary Fund International Monetary Fund, Washington, DC USA. Contact information at EDIRC.
Bibliographic data for series maintained by Akshay Modi ().