Mitigating housing market shocks: an agent-based reinforcement learning approach with implications for real-time...

Area of Science:

• Computational Economics
• Agent-Based Modeling
• Machine Learning in Economics

Background:

• Agent-based models (ABMs) are increasingly used for housing market dynamics due to data availability.
• Existing ABMs often lack adaptive decision-making capabilities for economic shocks.
• Machine learning (ML) offers potential enhancements for ABM frameworks.

Purpose of the Study:

• To investigate the impact of exogenous shocks on the UK housing market using an ABM.
• To integrate reinforcement learning (RL) for adaptive agent behavior within the ABM.
• To demonstrate emergent adaptive strategies in response to market volatility.

Main Methods:

• Development of an agent-based model (ABM) for the UK housing market.
• Integration of a reinforcement learning (RL) agent to control mortgage interest rates.
• Simulation of exogenous shocks to observe market dynamics and agent responses.

Main Results:

• Agents successfully learned real-time market trends and adapted decision-making to manage economic shocks.
• The model achieved objectives such as adjusting median house prices without predefined rules.
• The central bank agent exhibited emergent conservative behavior in sensitive scenarios, aligning with prior research.

Conclusions:

• Reinforcement learning effectively enhances ABM adaptability to housing market shocks.
• The developed model demonstrates emergent agent behaviors and policy adjustments.
• The approach is transferable to other complex housing market simulations.