Impact of Climate Water Risk on Corporate Operational and Capital Markets Performance: A Machine Learning Approach
Tian, Mingyan
2023
Abstract
Climate change and water availability impacts on corporate operational performance pose substantial risks to investors, shareholders, and the broader capital markets. Financial risks associated with climate and water are linked to short or long-term opportunity costs that are not disclosed in corporate accounting and formed the basis for the Task Force on Climate-Related Disclosures (TCFD), which forces corporations to disclose their financial risk exposures. The disclosure regulation aims to incentivize investment in corporate climate resilience through stewardship of natural resources. Given the knowledge gaps in water risk disclosure, quantitative approaches were developed to understand the financial materiality of water risk to corporate accounting and market performance. An exploration was conducted to test the hypothesis regarding the pricing of corporate water use intensity in the market and the potential quantification of this price premium using statistical tools and machine learning approaches. Indicators including water intensity relative to revenue, operating profit and net fixed assets were evaluated for representative companies from nine industry sectors. Using the statistical inference tool, propensity score matching (PSM), the analysis delved into the connection between water intensity and market metrics, accounting for corporate fundamentals. It showed that low water intensity results in improved returns over the benchmark (alpha), return on equity and long-term valuation (Tobin’s Q). In addition, water intensity based on corporate classification based on its activity was shown to be a poor proxy for water intensity benefits tied to financial metrics. The next step was to develop and test an imputation methodology combining econometric models and machine learning techniques to predict water intensity metrics for companies that are not disclosing water use risks. This methodology includes recursive feature elimination (RFE) method for feature selection, and the development of factor models using linear regression (OLS), generalized linear model (GLM), Lasso (LASSO), Random Forest (RF), and Adaptive boosting model (ADA). Random Forest models yielded the highest accuracy to impute water intensity indicators standardized to sales, operating profit and fixed assets from financial fundamentals, and allowed me to expand my testing universe from 500 to 2,525 company-years. Then, it explores the impact of water use indicators on market metrics, including share price return, short term operational (return on assets, ROA) and financial (return on equity, ROE) metrics, as well as long-term corporate valuation (using Tobin’s Q as a proxy). The difference between high- and low-water dependent companies, disclosing and non-disclosing (using imputed data) companies, as well as the impact of TCFD promulgation (2017) was tested. The results show that markets are rewarding companies exhibiting high water intensities with higher returns, though the effect is attenuated after TCFD implementation. Water intensity relative to sales and operating profit have a positive correlation with ROA and ROE, and a negative correlation to long-term value. Again, the coefficients for ROA and ROE are decreasing post-TCFD, while those for Tobin’s Q are increasing. Taken together, empirical evidence shows that markets are starting to price in water risk to companies. Interestingly, water intensity normalized to fixed asset investments exhibits a negative correlation to share price returns, indicating that investors are worried about capital-intensive companies delivering reduced returns. Using data science tools, my research offers new and valuable insights for business strategy and financial decision-making, emphasizing the need for managers to explore effective corporate water strategies to sustain or enhance competitiveness.Deep Blue DOI
Subjects
Water Intensity Financial Performance Machine Learning Water Strategy
Types
Thesis
Metadata
Show full item recordCollections
Remediation of Harmful Language
The University of Michigan Library aims to describe library materials in a way that respects the people and communities who create, use, and are represented in our collections. Report harmful or offensive language in catalog records, finding aids, or elsewhere in our collections anonymously through our metadata feedback form. More information at Remediation of Harmful Language.
Accessibility
If you are unable to use this file in its current format, please select the Contact Us link and we can modify it to make it more accessible to you.