ESG transparency in Mexico: a quantitative assessment of corporate reports using Natural Language Processing (NLP)

Abstract

This paper presents a methodology for measuring the presence of ESG criteria in the corporate information of Mexican public companies included in the Sustainable Corporate Accounting Index (SCI) using Natural Language Processing (NLP) techniques. Based on a corpus of 383 quarterly reports from 16 issuers included in the SCI from 2018 to 2023, a textual analysis process is developed: extraction, preprocessing, and linguistic analysis to generate quantitative metrics. Specifically, two scores are developed: the first measures the frequency and polarity of mentions related to sustainability, and a weighted risk score, which combines textual indicators of corporate risk (60%) with a measure of financial vulnerability based on the debt-to-EBITDA ratio (40%). The results indicate a lack of transparency and explicit reporting on companies' ESG practices.