| 主讲人简介： | Professor Danyang Xie, Chair Professor of the Thrust of Innovation, Policy, and Entrepreneurship, the Society Hub, HKUST (Guangzhou); Chair Professor of Economics, HKUST. Between 2019 and 2023, he served as the Acting Dean and Founding Dean of the Society Hub, HKUST (GZ).
Danyang Xie received his Ph.D in Economics from the University of Chicago in 1992. His previous affiliations include University of Montreal (Assistant Professor, 1991-1994), the International Monetary Fund (Economist and Senior Economist, 2000-2004), Tsinghua University (Special Appointment Professor, 2002-2005), and Wuhan University (Dean of the Economics and Management School, 2014-2017). At HKUST (1993-2002, 2004-2024), he served as Department Head in Economics between 2007 and 2010 and Director of Shui On Centre for China Business and Management between 2011 and 2013.
In addition, Professor Xie has been serving as a financial consultant expert for the 14th Standing Committee of the Guangdong Provincial People's Congress and the Deputy Director of the Sustainable Development Management Professional Committee of the China Management Science Society since 2023.
His research interests include economic growth and development, macroeconomics, digital economy, innovation and entrepreneurship, as well as money and banking. |
| 讲座简介： | This paper pioneers a novel approach to economic and public policy analysis by leveraging multiple Large Language Models (LLMs) as heterogeneous artificial economic agents. We first evaluate five LLMs' analytical capabilities in solving two-period consumption allocation problems under two distinct scenarios. In the first experiment, LLMs are given an explicit utility function, allowing us to rank them based on their rational optimization abilities. In the second experiment, LLMs are asked to make consumption-saving decisions purely based on their "gut feelings," revealing autonomous reasoning and intuition that not only aligns with established economic theories like the Life Cycle Hypothesis but also demonstrates diverse motivations including cultural factors and life-stage considerations. We then construct a Multi-LLM Agent-Based (MLAB) Model by mapping these LLMs to different income brackets in a calibrated population of 100 individuals. Using interest-income taxation as a case study, we demonstrate how MLAB Models can simulate policy impacts across heterogeneous agents. This methodology offers a promising new direction for public policy analysis by leveraging LLMs' unique advantages: their human-like reasoning capabilities, economic intuition, creative problem-solving abilities, and computational power, complementing traditional economic modeling approaches. |
