Predictive learning of operational water intensity from corporate financial indicators: Implications for AI-based resource management

Climate change is affecting water resource availability and predictability, impacting corporate operations, supply chains, and their financial performance or valuations in the capital markets. Scant corporate disclosure on water demand and operational intensity necessitates the use of advanced data science tools to inform risk exposures and investment decisions. Using 2550 company years across eleven industry sectors, machine learning models were built from financial data to quantify corporate water intensity metrics. These metrics were benchmarked to revenue, operating profit and investment in fixed assets. Fixed asset turnover, financial leverage, and inventory turnover were key predictors in factor models, particularly for production-oriented sectors such as IT, or consumer staples with R2 values from 0.66 to 0.75. Comparison of predicted water intensity data to disclosed information using global and industry-specific models indicated statistical agreement for selected sectors. When combined with text-based data, these insights inform firm-level trends of climate-water risk for financial resource allocations.