![[2510.12272] Heterogeneous RBCs via Deep Multi-Agent Reinforcement Learning](https://arxiv.org/static/browse/0.3.4/images/arxiv-logo-fb.png){kind=logo}
[2510.12272] Heterogeneous RBCs via Deep Multi-Agent Reinforcement Learning
Summary
This paper presents MARL-BC, a framework that combines deep multi-agent reinforcement learning with real business cycle models, addressing limitations in traditional economic modeling approaches.
Why It Matters
The integration of MARL with economic models allows for more flexible and realistic simulations of agent behavior, overcoming the computational burdens of existing heterogeneous-agent models. This advancement could significantly enhance the understanding of macroeconomic dynamics and improve policy-making.
Key Takeaways
- MARL-BC framework effectively combines multi-agent reinforcement learning with real business cycle models.
- It can recover traditional economic results while simulating rich agent heterogeneity.
- This approach addresses the computational challenges of existing heterogeneous-agent models.
Computer Science > Multiagent Systems arXiv:2510.12272 (cs) [Submitted on 14 Oct 2025 (v1), last revised 16 Feb 2026 (this version, v2)] Title:Heterogeneous RBCs via Deep Multi-Agent Reinforcement Learning Authors:Federico Gabriele, Aldo Glielmo, Marco Taboga View a PDF of the paper titled Heterogeneous RBCs via Deep Multi-Agent Reinforcement Learning, by Federico Gabriele and 2 other authors View PDF HTML (experimental) Abstract:Current macroeconomic models with agent heterogeneity can be broadly divided into two main groups. Heterogeneous-agent general equilibrium (GE) models, such as those based on Heterogeneous Agent New Keynesian (HANK) or Krusell-Smith (KS) approaches, rely on GE and 'rational expectations', somewhat unrealistic assumptions that make the models very computationally cumbersome, which in turn limits the amount of heterogeneity that can be modelled. In contrast, agent-based models (ABMs) can flexibly encompass a large number of arbitrarily heterogeneous agents, but typically require the specification of explicit behavioural rules, which can lead to a lengthy trial-and-error model-development process. To address these limitations, we introduce MARL-BC, a framework that integrates deep multi-agent reinforcement learning (MARL) with real business cycle (RBC) models. We demonstrate that MARL-BC can: (1) recover textbook RBC results when using a single agent; (2) recover the results of the mean-field KS model using a large number of identical agents; and (3)...
