Machine learning the macroeconomic effects of financial shocks

Niko Hauzenberger; Florian Huber; Karin Klieber; Massimiliano Marcellino

doi:10.1016/j.econlet.2025.112260

Machine learning the macroeconomic effects of financial shocks

Niko Hauzenberger, Florian Huber, Karin Klieber^*, Massimiliano Marcellino

^*Corresponding author for this work

Economics

Research output: Contribution to journal › Article › peer-review

Abstract

We propose a method to learn the nonlinear impulse responses to structural shocks using neural networks, and apply it to uncover the effects of US financial shocks. The results reveal substantial asymmetries with respect to the sign of the shock. Adverse financial shocks have powerful effects on the US economy, while benign shocks trigger much smaller reactions. Instead, with respect to the size of the shocks, we find no discernible asymmetries.

Original language	English
Article number	112260
Journal	Economics Letters
Volume	250
Early online date	14 Mar 2025
DOIs	https://doi.org/10.1016/j.econlet.2025.112260
Publication status	Published - Apr 2025

Funding

The authors thank Serena Ng for helpful comments and suggestions. Hauzenberger and Huber gratefully acknowledge financial support from the Jubil\u00E4umsfonds of the Oesterreichische Nationalbank (OeNB, grant no. 18763 ) and the Austrian Science Fund (FWF, grant no. ZK 35 ). Marcellino thanks for funding the European Union - NextGenerationEU, Mission 4, Component 2, in the framework of the GRINS -Growing Resilient, INclusive and Sustainable project (GRINS PE00000018 \u2013 CUP B43C22000760006). The views expressed in this paper do not necessarily reflect those of the Oesterreichische Nationalbank or the Eurosystem or the European Union, and they cannot be held responsible for them.

Keywords

Bayesian neural networks
Nonlinear local projections
Financial shocks
Asymmetric shock transmission

Access to Document

10.1016/j.econlet.2025.112260Licence: CC BY 4.0

Hauzenberger-etal-EconLet-Machine-learning-the-macroeconomic-effects-of-financial-shocksFinal published version, 1.77 MBLicence: CC BY 4.0

Cite this

@article{1ab83e3a9a424752800bd2b525bb58d3,

title = "Machine learning the macroeconomic effects of financial shocks",

abstract = "We propose a method to learn the nonlinear impulse responses to structural shocks using neural networks, and apply it to uncover the effects of US financial shocks. The results reveal substantial asymmetries with respect to the sign of the shock. Adverse financial shocks have powerful effects on the US economy, while benign shocks trigger much smaller reactions. Instead, with respect to the size of the shocks, we find no discernible asymmetries.",

keywords = "Bayesian neural networks, Nonlinear local projections, Financial shocks, Asymmetric shock transmission",

author = "Niko Hauzenberger and Florian Huber and Karin Klieber and Massimiliano Marcellino",

year = "2025",

month = apr,

doi = "10.1016/j.econlet.2025.112260",

language = "English",

volume = "250",

journal = "Economics Letters",

issn = "0165-1765",

publisher = "Elsevier B.V.",

}

TY - JOUR

T1 - Machine learning the macroeconomic effects of financial shocks

AU - Hauzenberger, Niko

AU - Huber, Florian

AU - Klieber, Karin

AU - Marcellino, Massimiliano

PY - 2025/4

Y1 - 2025/4

N2 - We propose a method to learn the nonlinear impulse responses to structural shocks using neural networks, and apply it to uncover the effects of US financial shocks. The results reveal substantial asymmetries with respect to the sign of the shock. Adverse financial shocks have powerful effects on the US economy, while benign shocks trigger much smaller reactions. Instead, with respect to the size of the shocks, we find no discernible asymmetries.

AB - We propose a method to learn the nonlinear impulse responses to structural shocks using neural networks, and apply it to uncover the effects of US financial shocks. The results reveal substantial asymmetries with respect to the sign of the shock. Adverse financial shocks have powerful effects on the US economy, while benign shocks trigger much smaller reactions. Instead, with respect to the size of the shocks, we find no discernible asymmetries.

KW - Bayesian neural networks

KW - Nonlinear local projections

KW - Financial shocks

KW - Asymmetric shock transmission

U2 - 10.1016/j.econlet.2025.112260

DO - 10.1016/j.econlet.2025.112260

M3 - Article

SN - 0165-1765

VL - 250

JO - Economics Letters

JF - Economics Letters

M1 - 112260

ER -

Machine learning the macroeconomic effects of financial shocks

Abstract

Funding

Keywords

Access to Document

Fingerprint

Cite this