Can AI and Econometrics Improve Climate Risk Management?

Climate change is increasingly recognised as one of the most significant sources of long-term financial risk facing the global economy. Over the last decade, the discussion has shifted beyond environmental sustainability towards a broader question of financial stability, with central banks, financial regulators, institutional investors, and policymakers seeking to understand how climate-related risks affect asset prices, corporate performance, and the resilience of financial systems. This shift is reflected in the growing body of research on climate finance, including the work of Patrick Bolton, Marcin Kacperczyk, Stefano Giglio, Robert Engle, and Bryan Kelly, whose studies have demonstrated that climate risk is increasingly priced by financial markets and should be considered alongside more traditional sources of financial risk.

The literature generally distinguishes between physical and transition risks.

• Physical risks arise from the direct economic consequences of climate change, including floods, hurricanes, droughts, heatwaves, and wildfires, all of which can disrupt production, damage infrastructure, reduce property values, and increase firms' financing costs.
• Transition rusks or the so called "carbon risk", by contrast, reflect the adjustments associated with moving towards a lower-carbon economy. Changes in enviromental regulation, carbon pricing, technological innovation, investor preferences, and corporate disclosure requirements can all alter firms' expected profatibility and market valuation. As highlighted by the Network for Greening the Financial System (NGFS), both forms of risk have high potential to generate systemic financial consequences if they are not appropriately identified and managed.

An important insight emerging from recent research is that climate risk is not priced uniformly across firms or regions. Engle et al. (2020), for example, demonsteate that climate-related news contains valuable information for financial markets and can be incorporated into dynamic hedge portfolios, while Giglio, Kelly (2020) and co-authors show that investors increasingly demand compensation for exposure to long-term climate uncertainty. Similarly, Bolton and Kacperczyk (2021) find that firms with higher carbon emissions face higher expected stock returns, suggesting that investors require an additional premium to hold carbon-intensive assets. Together, these studies indicate that climate risk has become a measurable financial factor rather than simply an enviromental concern. This growing recognition has created new challenges for financial institutions. Traditional financial statements alone provide only a partial picture of climate exposure, particularly as many risks have a strong spatial dimension. The financial consequences of climate change often depend on where assets are located, how vulnerable local infrastructure is to extreme weather events, and how regional climate policies evolve over time. As a result, investors increasingly require analytical approaches capable of integrating environmental information with conventional financial data.

Recent advances in artificial intelligence, remote sensing, and econometric modelling offer promising opportunities to address many of these challenges. Satellite imagery, geospatial datasets, climate indicators, and textual information extracted from financial news can now be analysed using machine learning techniques to monitor environmental change at an unprecedented scale. When combined with traditional econometric methods, these data allow researchers to identify more robust relationships between climate events and financial outcomes, improve climate risk forecasting, and develop more accurate measures of firms' and investors' exposure to both physical and transition risks.

Rather than replacing established quantitative methods, artificial intelligence is expanding the analytical toolkit available to economists and financial practitioners. Machine learning excels at processing large, high-dimensional datasets and uncovering complex patterns that may not be immediately apparent through conventional statistical techniques. Econometric models, however, remain essential for establishing causal relationships, testing economic hypotheses, and producing transparent evidence capable of supporting investment decisions, financial regulation, and public policy. Increasingly, the future of climate risk analysis lies in the integration of these complementary approaches rather than in choosing one over the other.

The growing importance of this interdisciplinary field is reflected in the increasing attention it receives from both academia and industry. For example, the upcoming Lisbon Sustainability Week 2026, hosted by CATÓLICA-LISBON School of Business & Economics under the theme Building Sustainability Together, brings together leading academics, policymakers, investors, regulators, and business leaders to discuss the future of sustainable finance, climate economics, and responsible investment. Among the keynote speakers is Nobel Laureate Robert F. Engle, whose pioneering work on financial volatility and, more recently, on climate risk and financial markets has helped shape the emerging field of climate finance. His research demonstrates how climate-related news and advanced econometric techniques can be used to construct dynamic hedge portfolios and improve the measurement of climate-related financial risk. The prominence of these discussions at major international forums highlights a broader shift: climate risk management is no longer viewed as a niche area of sustainable investing but as a core component of modern financial economics and risk management.

Although significant advances have been made in both climate finance and artificial intelligence, these developments have largely progressed along separate research paths. On the one hand, economists and financial researchers have made considerable progress in measuring how climate-related risks influence asset prices, corporate valuations, and portfolio performance. On the other hand, rapid developments in artificial intelligence, remote sensing, and geospatial analytics have transformed the ability to monitor environmental change with increasingly high spatial and temporal resolution.

Despite these parallel advances, relatively few studies have successfully integrated these two fields. Climate risk assessments often rely on financial statements, ESG metrics, or macroeconomic indicators, while AI-based remote sensing research typically focuses on detecting environmental hazards such as flooding, wildfires, coastal erosion, or land-use change without explicitly linking these observations to financial outcomes. As a result, valuable environmental intelligence is often disconnected from the investment decisions and risk management frameworks used by financial institutions.

Climate risk has evolved from being viewed as a sustainability issue to becoming a central component of modern financial risk management. As the frequency and economic impact of extreme weather events continue to increase, investors, regulators, insurers, and policymakers face growing pressure to develop more sophisticated methods for identifying, measuring, and managing climate-related financial risks. Throughout this article, we have argued that neither traditional financial models nor artificial intelligence alone are sufficient to meet this challenge. Instead, the future of climate finance lies in combining the predictive capabilities of AI with the statistical rigour and interpretability of econometric methods.

Recent advances in remote sensing, machine learning, and geospatial analytics are transforming the quantity and quality of information available for climate risk assessment. At the same time, econometric techniques remain indispensable for understanding causal relationships, testing hypotheses, and translating complex environmental signals into robust financial evidence. The integration of these complementary approaches offers the potential to produce more accurate forecasts, strengthen climate stress testing, improve portfolio allocation, and support evidence-based financial regulation.

Looking ahead, several research priorities remain. Greater attention should be given to the development of hybrid AI-econometric models that combine predictive performance with transparency and explainability. Future work should also focus on integrating multiple sources of information, including satellite imagery, climate projections, ESG disclosures, financial news, and socioeconomic indicators, into unified modelling frameworks capable of capturing both physical and transition risks. Equally important is ensuring that these models account for regional differences, behavioural responses, and social vulnerability so that climate risk assessments remain both scientifically robust and relevant for public policy.

Many of these ideas build upon the pioneering work of Nobel Laureate Robert F. Engle, whose research has fundamentally changed how economists think about financial risk. His contributions extend beyond the development of volatility models to demonstrate that climate-related information itself can be treated as a measurable financial risk factor. By combining textual analysis, financial markets, and advanced econometric techniques, Engle's work on climate news hedging has opened new directions for research into sustainable finance and has helped establish climate risk as an integral part of modern asset pricing and portfolio management. As discussions at international forums such as Lisbon Sustainability Week continue to demonstrate, the intersection of AI, econometrics, and climate finance is becoming one of the most dynamic areas of contemporary economic research.

As these disciplines continue to converge, demand is growing for professionals capable of combining economic intuition with advanced quantitative skills. At Timberlake Consultants, we are committed to supporting this transition through specialised training in econometrics, artificial intelligence, machine learning, data science, and sustainable finance. Our upcoming courses are designed to equip researchers, practitioners, policymakers, and financial professionals with the tools needed to analyse increasingly complex datasets and develop robust evidence for decision-making. By bridging academic research with practical application, these programmes aim to prepare the next generation of economists and analysts to address one of the defining financial challenges of the twenty-first century.