核心概念
TaxAI is a realistic economic simulator that showcases the effectiveness of MARL algorithms in optimizing tax policies and household strategies.
要約
Taxation and government spending are vital tools for economic growth and social equity.
Multi-Agent Reinforcement Learning (MARL) is proposed as a solution to model dynamic games between the government and households.
TaxAI, based on the Bewley-Aiyagari economic model, benchmarks traditional methods against MARL algorithms.
Results show MARL's superiority in optimizing tax policies and revealing tax evasion behavior among households.
The scalability of TaxAI allows simulations with up to 10,000 agents, surpassing traditional methods.
統計
MARLアルゴリズムは最適な税制策定に優れている。
TaxAIはBewley-Aiyagari経済モデルに基づいている。
MARLアルゴリズムは従来の方法よりも優れている。