Can weather events and climate conditions help predict the chance a borrower defaults on their mortgage? A recent paper by Viani Djeundje Biatat, Jonathan Crook and Galina Andreeva finds that they can.

In July 2026, Jonathan delivered a plenary presentation at the Mathematics in Finance conference, jointly organised by the University of Cape Town, North West University and the University of Pretoria, and attended by more than 100 delegates from all corners of the world.

The presentation showed that model accuracy improves when a range of weather variables and flexible statistical relationships are included, and that omitting weather predictors in regions such as California can lead banks to underestimate the capital they need to hold against severe events.

Jonathan also presented a second paper from the same research team, on a methodology for predicting a mortgage's expected future cash flow at any point in its life, incorporating anticipated changes in the macroeconomy. The method adjusts expected cash flow as an account holder moves between different states of delinquency, even where they do not ultimately default.

References

Djeundje, V., Crook, J.N. and Andreeva, G. (2026) "Disentangling Environmental and Economic Risks in Mortgage Default: A Spatio-Temporal Modelling and Stress-Testing Framework with Capital Implications." Available at SSRN

Djeundje, V., Crook, J. and Andreeva, G. (2025) "The Devil in the Details: Dynamic Prediction of Loan Profitability with Macroeconomic Drivers through Multistate Modelling." European Journal of Operational Research, 327(2), 703-715.

On 13 July 2026, the Federal Reserve Bank of Philadelphia and the Supervisory Research Forum (SURF) hosted the Workshop on Innovations in Credit Scoring. This full-day virtual event looked at how AI, machine learning and alternative data are changing credit scoring, and what that means for transparency, data privacy, risk management and access to credit.

The workshop's two academic advisors were Professor Stefan Lessmann of Humboldt University of Berlin, an external affiliate of the Credit Research Centre, and Galina Andreeva, Professor of Societal Aspects of Credit and Director of the Credit Research Centre.

Galina also chaired the panel “The Future of Credit Scoring”, which brought together Kiran Aware of Happen Bank, Rikard Bandebo of VantageScore, Professor Cristián Bravo of Western University, Ethan Dornhelm of FICO, Dmytro Kolechko of Wing Bank, Stefan Lessmann and Dan Richard of Prism Data.

In the first module, Stefan presented “Beyond Traditional Scoring: Harnessing Advances in AI to Improve Predictive Accuracy”, and he went on to chair the second module. Dmytro Kolechko, CEO of Wing Bank, presented “Digital Footprint Credit Scoring Modeling”, the paper that won the Best Paper Award at the Centre’s Credit Scoring and Credit Control Conference in 2025.

The programme also included a keynote by Catherine Tucker and research presentations on AI, machine learning and alternative data from academia and industry.

In June and July 2026, Viani Djeundje Biatat of the Credit Research Centre, together with Marc Gaudart and Paul Randall of Trent Advisory Services, delivered a specialised programme on credit scoring and digital decision-making for financial institutions and central bank staff across the West African Economic and Monetary Union. It ran online on 24 and 29 June, then in person in Dakar, Senegal, from 3 to 10 July. The programme was organised by COFEB, the banking training centre of the Central Bank of West African States (BCEAO), with support from the Centre.

Around 400 applications were received and 183 took part. Participant feedback highlighted the quality of the teaching, the relevance of the topics covered and the practical value of the training; planning for future editions is already in progress.

The programme builds on Viani’s work supporting risk analytics capability across Africa since 2017, including “Building Africa’s future in credit risk and analytics”, delivered at the African Institute for Mathematical Sciences in Cameroon in December 2024.

With thanks to the International Centre for Mathematical Sciences (ICMS) for its support.

On 26 June 2026, Galina Andreeva, Director of the Credit Research Centre, delivered a keynote at the 3rd Annual Credit Risk Model Conference, held in Bangkok.

Her keynote, "The Credit Research Centre: Era of AI, Regulation, and Societal Scrutiny," explored the evolving landscape of credit risk modelling at a time when artificial intelligence, regulatory expectations and societal accountability are converging.

After an introduction from Professor Raffaella Calabrese, Benjamin Guin of the Bank of England spoke on green mortgage innovation, and his colleague Mahmoud Fatouh on how biodiversity interacts with default risk in the UK mortgage market. Scott Harrison of Experian looked at how banks can source and act on climate and nature risk data, Douglas Phillips of Arup gave an overview of the Taskforce on Nature-related Financial Disclosures, and Christopher Sterley of True North Partners spoke on advancing climate risk models to capture nature risk. Viani Djeundje Biatat of the Credit Research Centre presented research on wildfires, weather, mortgage defaults and capital.

The afternoon closed with a panel discussion and a summary from Peter Baker of Edinburgh Innovations.

The editors describe the discipline as being at "a critical juncture," shaped by fast-moving technology, tightening regulation, and growing scrutiny of algorithmic decisions.

Automated credit scoring is now standard practice across retail and corporate lending in the financial industry, a direct application of decades of quantitative methods research. This special issue continues that tradition, advancing knowledge and application at a time when that scrutiny and regulatory pressure are continuously increasing.

Edited by Galina Andreeva (Edinburgh), Jonathan Crook (Edinburgh) and Christophe Mues (Southampton), and published in May 2026 in the Annals of Operations Research, the issue is linked to the Centre's Credit Scoring and Credit Control conferences, held since 1989.

The contributions span four areas. On stress-testing and loss modelling, papers examine how transition matrix choices shape long-run default projections, extend stress-testing methods to IFRS 9 overlays, and propose new ways to estimate loss and profitability in online lending. On machine learning and alternative data, authors draw on sources as varied as social media, ESG reports and satellite imagery to strengthen credit rating and scoring models. Further research tackles practical concerns beyond default prediction, from reject inference to fraud detection to model monitoring.

Finally, a standout paper turns to algorithmic fairness, examining how credit models can amplify inequity across intersecting characteristics such as gender, age and marital status. A question of growing importance as automated decisions play a larger role in who gets access to credit.

Read the full special issue on SpringerLink

As background, the price cap means that when HCSTC loans are taken out or rolled over, the combined interest and fees charged must not exceed 0.8% per day of the amount borrowed. If borrowers default, fees must not exceed £15. Firms can continue to charge interest after default but not above the initial rate. Finally, borrowers must never have to pay more in fees and interest than 100% of what they borrowed.

The price-cap was last reviewed in 2017. Since then, the market and wider economic environment have changed significantly, both for consumers and firms.

In 2017, the FCA made the decision to maintain the HCSTC price cap at its current level and committed to looking at it again after three years but chose to delay because of Covid and the cost-of-living challenges.

This Roundtable offered a key opportunity to consider whether the price cap continues to work effectively in the interests of consumers and markets and remains fit for purpose in the current economic environment.

The course has been designed specifically for financial institutions and is led the CRC’s own Dr Raffaella Calabrese, and Dr Joseph Breeden, CRC External Affiliate, CEO of Deep Future Analytics LLC, and President of the Model Risk Managers’ International Assocation.

About the Course

Participants will gain both the conceptual foundations and practical tools to assess the financial implications of climate change for credit risk. The course covers how to select and apply appropriate climate scenarios, how to integrate climate-related risks into existing credit risk frameworks, and how to evaluate the resilience of loan portfolios under severe stress conditions. It is designed to be tailored to the specific needs of individual organisations, and prospective participants are encouraged to contact the Business School to discuss collaboration opportunities.

The course is particularly well suited to credit risk modellers, senior risk managers, and financial professionals seeking to build or strengthen their organisation's capacity in this rapidly evolving area.

Learn more & register

The research behind the course

Climate Stress Testing and Climate Scenarios draws on a substantial and growing body of research on stress-testing developed by the CRC with Raffaella and Joe making meaningful contributions (please see the details below). State-of-the art stress-testing approaches developed for macroeconomic impacts have found a novel and profound application in addressing climate change issues and their estimations.

The climate applications continue a long-standing legacy of stress‑testing methodologies advanced by CRC founder, Professor Emeritus Jonathan Crook, and other CRC members and External Affiliates:

• Djeundje & Crook (2025), Incorporating behavioural and macroeconomic correlations for the prediction of bank capital for credit risk.
• Wang, Crook, Andreeva (2025), Sensitivity of model parameter estimates in stress testing the probability of default: evidence over the financial crisis.
• Bocchio, Crook &Andreeva (2023), The impact of macroeconomic scenarios on recurrent delinquency: A stress testing framework of multi-state models for mortgages.
• Djeundje & Crook (2022), Sensitivity of stress testing metrics to estimation risk, account behaviour and volatility for credit defaults.
• Wang, Crook & Andreeva (2020), Reducing estimation risk using a Bayesian posterior distribution approach: Application to stress testing mortgage loan default.
• Bellotti & Crook (2014), Reducing estimation risk using a Bayesian posterior distribution approach: Application to stress testing mortgage loan default.
• Bellotti & Crook (2013), Forecasting and stress testing credit card default using dynamic models.

By grounding learning in proven stress‑testing methodologies, the course enables participants to understand not only how climate scenarios are constructed, but also how they can be embedded in empirical models relevant for financial decision‑making through example from Raffaella’s and Joe’s work.

Dr Raffaella Calabrese's research focuses on the quantitative modelling of climate risk in lending and mortgage markets. Two recent studies are of particular relevance to the course. Her work on the impacts of extreme weather events on mortgage risks in Florida — examining how hurricane and flood exposure affects default behaviour and how those relationships evolve under projected climate change scenarios — provides a compelling empirical foundation for understanding physical climate risk at the asset level. A complementary study on climate stress testing for mortgage default probability develops methodological tools for translating macro-level climate scenarios into granular credit risk estimates, offering practitioners a direct pathway from scenario design to portfolio-level impact assessment. Raffaella also designs and delivers executive courses for the European Central Bank on Climate Stress Testing and the EU Supervisory Digital Finance Academy.

Dr Joseph Breeden's contributions bring a complementary focus on the practical dynamics of climate stress in lending. His 2023, Impacts of Drought on Loan Repayment, examines how prolonged drought conditions affect borrower repayment behaviour, drawing on empirical data to quantify a risk that is increasingly material for agricultural and regional lenders, but whose methods and findings carry wider lessons for physical risk modelling.

A second paper, co-authored with Y. Leonova and A. Bellotti, examines instabilities in the use of Cox Proportional Hazards models for loan portfolio forecasting and stress testing — a technically important contribution that speaks directly to the reliability of the modelling approaches practitioners use when assessing portfolio resilience under adverse scenarios.

A collaboration rooted in complementary expertise

The partnership between the CRC and B-CCaS represents a natural alignment of strengths. The CRC brings deep expertise in credit risk analytics, scorecard development, and financial modelling, while B-CCaS contributes specialist knowledge in climate science, sustainability strategy, and the policy dimensions of climate-related financial risk. The joint development of this course reflects the University of Edinburgh Business School's ambition to translate interdisciplinary research into professional practice.

The CRC looks forward to supporting financial institutions in building the analytical capabilities they need to navigate the risks and responsibilities that climate change presents to the credit sector.

Course presenters

Raffaella Calabrese, Professor at the University of Edinburgh Business School

Dr Calabrese works closely with financial institutions and regulators to propose analytical solutions to better measure the effects of climate change on credit risk and how to perform climate stress testing exercises. Raffaella collaborates with financial institutions, government bodies and regulators in the UK, USA, China and multiple EU countries. She provides training on Climate Stress Testing at the European Central Bank and on digital finance at the European Commission.

Joe Breeden, Chief Executive Officer at Deep Future Analytics LLC

Dr Breeden has been designing and developing risk management systems for loan portfolios since 1996. He founded Deep Future Analytics in 2011, which focuses on portfolio and loan-level forecasting solutions for pricing, account management, stress testing, CECL/IFRS9, and AI monitoring; serving banks, credit unions, and finance companies. He also serves as President of Model Risk Managers’ International Association and on the board of Upgrade. Dr Breeden earned a PhD in Physics and published over 90 academic articles, 8 patents, and 6 books.