生成性经济建模（英）

BIS Working PapersNo 1312 Generative economic modeling by Hanno Kase, Matthias Rottner and Fabio Stohler Monetary and Economic Department December 2025 JEL classification: C11, C45, D31, E32, E52 Keywords: machine learning, neural networks, nonlinearities, heterogeneous agents BIS Working Papers are written by members of the Monetary and Economic Department of the Bank for International Settlements, and from time to time by other economists, and are published by the Bank. The papers are on subjects of topical interest and are technical in character. The views expressed in this publication are those of the authors and do not necessarily reflect the views of the BIS or its member central banks. This publication is available on the BIS website (www.bis.org). © Bank for International Settlements 2025. All rights reserved. Brief excerpts may be reproduced or translated provided the source is stated. ISSN 1020-0959 (print) ISSN 1682-7678 (online) Generative Economic Modeling*Hanno KaseEuropean Central BankMatthias RottnerBank for International SettlementsDeutsche BundesbankFabio StohlerUniversity of BonnDecember 2, 2025AbstractWe introduce a novel approach for solving quantitative economic models: generativeeconomic modeling. Our method combines neural networks with conventional solu-tion techniques. Specifically, we train neural networks on simplified versions of theeconomic model to approximate the complete model’s dynamic behavior. Relying onthese less complex submodels circumvents the curse of dimensionality, allowing the useof well-established numerical methods. We demonstrate our approach across settingswith analytical characterizations, nonlinear dynamics, and heterogeneous agents, em-ploying asset pricing and business cycle models. Finally, we solve a high-dimensionalHANK model with an occasionally binding financial friction to highlight how aggregaterisk amplifies the precautionary motive.Keywords— Machine Learning, Neural networks, Nonlinearities, Heterogeneous AgentsJEL codes— C11, C45, D31, E32, E52*hanno.kase@ecb.europa.eu, matthias.rottner@bis.org, fabio.stohler@uni-bonn.de.We would like tothank Christian Bayer, Jes´us Fern´andez-Villaverde, Alexandros Gilch, Keith Kuester, Jamie Lenney, LeonardoMelosi, Simon Scheidegger, and Yucheng Yang for their very helpful suggestions. We thank seminar par-ticipants at the Barcelona Summer Forum session on Machine Learning in Economics, the Conference onMachine Learning for Economics and Finance, Banque de France AI Methods Conference 2025, the Bank forInternational Settlements, the Dynare Conference 2025, and the ECONDAT 2025 Spring meeting for helpfulcomments and discussions. Fabio Stohler gratefully acknowledges support by the German Research Foun-dation (DFG) through CRC TR 224 (Project C05). The views in this paper are those of the authors andshould not be interpreted as reflecting the views of the Bank for International Settlements, the DeutscheBundesbank, the European Central Bank, or any oth