2025年12月18日(周四)13:30-15:00 邹杰 (山东大学)
讲座系列:现代财税理论与政策系列讲座第29场(学科发展系列讲座总期第109场)
讲座人:邹杰 (山东大学)
时间:2025年12月18日(周四)13:30-15:00
地点:博学楼1220
主持人:周波(对外经贸大学国际经济贸易学院财政学系教授)
讲座题目:A Machine-Learning Based Measure of Corporate Tax Avoidance
讲座摘要:We exploit a unique feature of the Chinese tax system: state-owned enterprises (SOEs) and private firms are subject to the same tax laws, yet SOEs face markedly weaker incentives to avoid taxes. Leveraging this contrast, we train a machine learning model to predict the profits of SOEs and apply it to non-SOE firms (public and private) to infer their actual profits — and thereby estimate the extent of their tax avoidance. Our measure aligns closely with empirical proxies for the costs and benefits of tax avoidance, consistent with the predictions of the classic Allingham-Sandmo tax evasion model. We further validate our approach by examining the effects of a nationwide digital tax administration reform, which acts as an exogenous shock to the cost of tax evasion. Lastly, we demonstrate that our measure is robust to concerns about accounting manipulation and structural differences in production between SOEs and private firms.
主讲人简介:
邹杰,山东大学经济学院副研究员,泰山学者青年专家,2023于清华大学经管学院获得经济学博士学位,曾为麻省理工大学访问学者。主要研究方向为:1)公共经济学,主要关注财政政策、税收政策与公共福利政策。2)公司金融,主要关注人工智能(机器学习与大语言模型)在公司金融问题中的应用。目前已经在《经济研究》、BAR等期刊发表论文多篇,所著论文曾获“第七次全国优秀财政理论研究成果奖一等奖”。主持国家自然科学基金青年项目、山东省自然科学基金青年项目,并参与多项国家级、省部级课题。