TaxAgent: How Large Language Model Designs Fiscal Policy

2025-06-03

AI Summary

Economic inequality exacerbates disparities in education, healthcare, and social stability, yet conventional tax frameworks—such as the U.S. federal income tax—and the Saez optimal taxation model fail to capture taxpayer heterogeneity and behavioral irrationality. This paper introduces an LLM-driven intelligent fiscal policy agent system, the first to deeply integrate large language models into agent-based macroeconomic modeling. We construct a co-simulation environment comprising heterogeneous household agents and an LLM-powered government policy agent, enabling dynamic, data-driven optimization of progressive tax rates. Our approach significantly outperforms the Saez model, the current U.S. tax regime, and a laissez-faire benchmark along the equity–efficiency trade-off frontier. Empirical results validate the feasibility, scalability, and interpretability of this novel paradigm for generating intelligent, adaptive fiscal policies grounded in microfoundations and behavioral realism.

Technology Category

• Machine Learning: Large Multimodal Models (LMMs)
• Search and Optimization: Learning to Search
• Multiagent Systems: Mechanism Design

Application Category

• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation
• Economics, Online Markets and Human Computation: Cost models of using LLMs in production systems

Abstract

Economic inequality is a global challenge, intensifying disparities in education, healthcare, and social stability. Traditional systems like the U.S. federal income tax reduce inequality but lack adaptability. Although models like the Saez Optimal Taxation adjust dynamically, they fail to address taxpayer heterogeneity and irrational behavior. This study introduces TaxAgent, a novel integration of large language models (LLMs) with agent-based modeling (ABM) to design adaptive tax policies. In our macroeconomic simulation, heterogeneous H-Agents (households) simulate real-world taxpayer behaviors while the TaxAgent (government) utilizes LLMs to iteratively optimize tax rates, balancing equity and productivity. Benchmarked against Saez Optimal Taxation, U.S. federal income taxes, and free markets, TaxAgent achieves superior equity-efficiency trade-offs. This research offers a novel taxation solution and a scalable, data-driven framework for fiscal policy evaluation.

Problem

• Addressing economic inequality through adaptive tax policies
• Overcoming limitations of traditional taxation systems and models
• Integrating LLMs with agent-based modeling for fiscal policy design

Innovation

• Integrates LLMs with agent-based modeling for tax policy
• Simulates heterogeneous taxpayers using H-Agents and TaxAgent
• Optimizes tax rates iteratively balancing equity and productivity