NSF Award
2419996
The project team leverages rich historical job-postings datasets to construct a new measure for organizations’ involvement in environment, social and governance (ESG)-related activities. In recent decades, there has been growing societal demand for organizations to attend to their ESG-related responsibilities. While organizations are under intense evaluation for their involvement in ESG, most of the popular ESG ratings used by both academics and the investor community are produced by third-party rating agencies. These ratings are largely a black box, with little transparency of the underlying input data or the methodologies that have been employed to produce these ratings. It is thus risky for the academic and the broader investor communities to over-rely on such third party-generated ratings due to the lack of accuracy, transparency, or reproducibility. As a result, constructing a new ESG measure with improved quality and transparency is desired, which motivates this project.<br/><br/>With the assumption that organizations need human capital to get the work done, jobs offered by organizations reveal important information about organizational priorities. This project leverages recent availability of 250 million job-postings from 2008 to 2023 (and are still being updated) for 60,000 companies of various sizes to compute a job-posting-based novel ESG measure. The team employs the pre-train and fine-tune paradigm of text representation learning, upon which a fine-tuned large-language model (LLM) coupled with a neural network-based classifier is developed and applied to the textual job postings for constructing the ESG measure. The project aims at delivering three sets of output: (1) a new measure of organizational ESG engagement (and relevant scores in sub-categories), (2) the generic framework and methodology underlying the construction of this new measure, and (3) a report and a re-evaluation study of a widely-cited research article on ESG. All three sets of products contribute to organizational research on ESG and will be shared with the public.<br/><br/>This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.
