时间：2026年5月20日（周三）15:30-17:00

地点：厦门大学嘉庚二号楼（成枫楼）512

演讲嘉宾：黄乃静  教授

演讲题目：From micro to macro: Learning real-time economic signals from firm-level accounting data

摘要：

How can real-time information about aggregate economic activity be extracted from firm-level accounting data? This paper proposes a micro-to-macro nowcasting approach that uses machine learning algorithms to directly exploit the full cross-section of accounting information from 21,061 publicly listed U.S. firms for nowcasting U.S. aggregate output. Rather than relying on aggregated corporate indicators, the proposed framework preserves the rich information contained in firm-level heterogeneity and cross-firm interactions that is often lost through aggregation. The empirical results show that the proposed approach significantly improves nowcasting accuracy, reducing the root mean squared error by more than 70% relative to a random walk benchmark and substantially outperforming models based on aggregated accounting data. Firm-level corporate accounting data also improve nowcasting performance by about 17.35% relative to models using mixed-frequency aggregate macroeconomic and financial predictors. Our feature-importance analysis shows that the firms with larger assets, lower book-to-market ratios, or lower leverage, and the firms in the industrials, utilities, and consumer discretionary sectors have greater importance for nowcasting U.S. GDP. Furthermore, although the number and composition of the firms that release quarterly financial reports differ over time before the end of a quarter, the GDP nowcasts remain quite stable. Overall, the results underscore the value of directly exploiting firm-level accounting data from all publicly listed firms for nowcasting aggregate economic activity。

主持人：刘峰  会计发展研究中心主任/会计学系教授

主办单位  厦门大学会计发展研究中心、厦门大学会计学系、厦门大学会计学系MPAcc中心

项目支持  冠亚厦门大学会计系发展基金