Corporate Sustainability: A Model Uncertainty Analysis of Materiality
本文复制了Khan等人关于企业可持续性重要性可预测股票回报的发现，但认为这是统计假象，并通过机器学习分析表明，基于历史数据准确判断可持续性哪些方面对投资者重要可能很困难。
ABSTRACT For decades, scholars searched for a connection between a corporation's current performance with respect to sustainability and the future returns of its stock. In 2016, Khan, Serafeim, and Yoon published an apparent breakthrough in this quest: guidance on materiality from the Sustainability Accounting Standards Board allowed the construction of corporate sustainability scales that reliably predicted stock returns. Their finding had immediate and broad impact, but it remains, in its authors' own words, just “first evidence.” Here, we further explore the relationship between material-sustainability and stock returns by performing a “model uncertainty analysis.” We reproduce the original estimate but conclude that it is a statistical artifact. We then use machine learning to explore the practicality of employing historical associations to determine which aspects of sustainability are material to investors. We conclude that, for one popular source of data on corporate sustainability, accurate guidance on materiality may be difficult to achieve. JEL Classifications: Q51; D22; L25; C11; C18.