ESG and Hong Kong real estate listed companies : a textual analysis

Abstract

This dissertation presents a thorough textual analysis of the annual reports from real estate companies in Hong Kong. The primary focus is Environmental, Social, and Governance (ESG) investing within the real estate sector. The impetus for this research lies in the discernible gap in comprehensive studies that pinpoint potential determinants of ESG disclosures within the real estate industry and Hong Kong's equity market. Initially, the analysis employs textual analysis techniques to examine the involvement of listed real estate companies with ESG disclosures. The primary focus of this textual analysis is the frequency of ESG-related words. The study uses a dictionary based on exhaustive words about ESG in 10-K filings in the United States (US), derived from previous literature. To tailor-fit to the context of Hong Kong and the real estate industry, there were interviews conducted with industry professionals to fine-tune the dictionary. This step was crucial in ensuring the dictionary's robustness for tracking ESG disclosures. Based on the advice during these interviews, Dictionary 2 was created, incorporating additional ESG words such as 'green certificates' beyond the original dictionary, thereby serving as a robustness test. Following these interviews, a textual analysis was carried out using Python and other libraries. The study computed the frequency and respective percentages of ESG words. Primarily, the investigation included 40 listed real estate companies, chosen based on their firm size and the readability of information available in Refinitiv Workspace and Bloomberg Terminal. The study period spans from 2004 to 2023. In addition, this data was subject to further processing and analysis. The results indicated a gradual year-by-year increase, with the 'Environment' pillar taking precedence over percentage changes over the other two pillars. Year-over-year percentages also displayed intriguing conclusions for periods when there were sudden surges in ESG word instances. The evidence highlights that the Hong Kong real estate industry is engaged with ESG opportunities. The study also designed ratios of ESG words to finance words to weed out noise and display the weighting between ESG topics and finance topics, the latter of which should be the mainstream in official reports issued by listed firms. Another aspect of the study explored the correlation between instances of ESG words, ESG performance scores from Refinitiv Workspace, and ESG disclosure scores from Bloomberg Terminal. The results show a moderately strong correlation between the percentage of ESG words, the ratio of ESG to finance words, and the two scores, suggesting a potential method for approximating third-party ESG scores using ESG word instances. A regression analysis was conducted to define the determinants of ESG instances. The model applied was the Estimated Generalised Least Square (GLS), which minimizes the effects of heteroscedasticity. Using ESG word percentage as the dependent variable, the results show that regulatory changes (post-2016), Hong Kong origin, investment property focus, and firm size significantly boost ESG word instances. Using ESG word percentage from Dictionary 2, the robustness test projected similar and even more resilient results. Finally, a placebo test was conducted using finance word percentage. Although the independent variable statistically significantly explained the dependent variable in the model, the model revealed insignificant explanatory power overall, hinting at the original regression model's moderate robustness. This research impacts understanding ESG disclosures in Hong Kong's real estate industry. It highlights vital determinants impacting such disclosures and offers a practical methodology for stakeholders to assess firms' ESG commitment. The findings are crucial for investors, regulators, and companies to refine their ESG processes, underlining the growing importance of ESG factors and the need for further research in this area.