European sovereign debt control through reinforcement learning

Area of Science:

• Economics
• Macroeconomic modeling
• Computational economics

Background:

• Economic system resilience depends on stakeholder coordination during shocks.
• Past crises (Eurozone, COVID-19) highlight coordination failures and their consequences.
• Asymmetric shocks within heterogeneous economic regions necessitate coordinated policy responses.

Purpose of the Study:

• To demonstrate the importance of cooperation in monetary and fiscal policies during emergencies.
• To address macroeconomic non-resilience, with a focus on public debt management.
• To develop and test a macroeconomic modeling framework for simulating cooperative policy interactions.

Main Methods:

• Development of a macroeconomic modeling framework.
• Integration of nonlinear control models with reinforcement learning techniques.
• Implementation of the Soft Actor-Critic algorithm for policy optimization.

Main Results:

• The Soft Actor-Critic algorithm optimizes policy responses for key macroeconomic variables.
• The algorithm demonstrates comparable or superior performance to Nonlinear Model Predictive Control (NMPC).
• Simulations show the effectiveness of cooperative regimes in managing macroeconomic stability.

Conclusions:

• Coordinated monetary and fiscal policies are crucial for economic resilience.
• Reinforcement learning, specifically Soft Actor-Critic, offers a powerful tool for macroeconomic policy optimization.
• The findings are particularly relevant for multi-actor economic regions like the Euro area facing asymmetric shocks.