The Role of AI in Economic Empirical Analysis: Focus on Large Language Models in Computational Economics

This article analyzes the integration of Large Language Models [LLMs] into economic empirical analysis, particularly in computational economics, exploring theoretical contributions, implementation challenges, and 1-5 year outlooks.​Theoretically, LLMs revolutionize economic modeling. They enable realistic agent-based simulations with diverse decision-making, automate economic relationship identification for improved causal inference, and integrate multi-modal data for better financial/economic forecasting. Additionally, LLMs capture behavioral nuances for personalized economic analysis.​ However, practical implementation is hindered by data limitations [availability, quality, privacy], model opacity, generalization issues, alignment with economic theory, and interdisciplinary expertise gaps.​ In the near future, specialized economic LLMs, enhanced agent-based frameworks, advanced causal inference methods, and real-time economic monitoring tools are expected. These will reshape policy analysis and data interpretation, though ethical and regulatory adaptations are needed to address bias and transparency.​In conclusion, LLMs hold transformative potential for computational economics, contingent on overcoming technical and institutional hurdles while integrating theory and practice.