Study of taxes, regulations and inequality using machine learning algorithms.

Area of Science:

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

Background:

• Genetic machine learning (ML) algorithms effectively train agents in economic models like the Yard-Sale model to optimize wealth.
• A key finding is that increased fractions of rational agents lead to greater collective inequality, diminishing trade and liquidity.
• Most agents, even with rational behavior, often end up with zero wealth.

Purpose of the Study:

• To investigate the impact of a taxation-redistribution mechanism on wealth inequality within ML-driven economic models.
• To analyze how rational agents' behavior interacts with regulatory interventions.
• To explore strategies for mitigating inequality while maintaining agent optimization.

Main Methods:

• Utilizing genetic machine learning (ML) algorithms to simulate agent behavior in the Yard-Sale model.
• Integrating a taxation-redistribution mechanism into the ML algorithm.
• Analyzing the effects of varying fractions of rational agents and redistribution parameters on wealth distribution.

Main Results:

• The introduction of a taxation-redistribution mechanism can significantly reduce inequality, particularly when agent behavior remains constant.
• Rational agents, while seeking optimal wealth, can select risk levels compatible with specific parameters, but this can still lead to increased inequality.
• Even with redistribution, achieving equitable wealth distribution remains a challenge, with some agents potentially still ending with minimal wealth.

Conclusions:

• Taxation-redistribution mechanisms are crucial for mitigating wealth inequality generated by rational agents in ML economic models.
• Balancing agent optimization with societal equity requires careful calibration of redistribution parameters.
• Further research is needed to fully reconcile individual wealth maximization with collective economic stability and fairness.