Innovations in credit risk assessment of small and medium enterprises

Pleshkevich, B. (2026) Innovations in credit risk assessment of small and medium enterprises. PhD thesis, University of Reading. doi: 10.48683/1926.00130663

Abstract/Summary

The credit risk assessment of small and medium enterprises (SMEs), as a separate class, has been specifically segmented due to their high dependence on external finance as well as critical importance for global economies. This thesis deepens understanding of tools and methods that can be utilized for credit risk modelling to reveal various new patterns in lending to SMEs with three essays. First essay summarizes the key developments in the SME credit risk modelling considering the modern estimation techniques. Through a literature review and an empirical study, it outlines the historical evolution of the SME credit risk evaluation and presents current trends to discuss recent contributions from utilization of advanced techniques for SME scoring such as higher performance and robustness. Second essay proceeds by investigating obstacles for quantitative assessment. By reviewing various definitions of credit risk events in application to multinational SME loans, this part demonstrates how utilization of comprehensive robustness specifications promotes capturing similarities across European countries through empirical study on Spanish and Italian SMEs. Finally, third expands the discussion by looking into Generative AI (GenAI) applications and their great potential to improve credit risk assessments for SMEs in Europe. Assessment of new capabilities of GenAI tools with regulatory and ethical considerations are used to propose a quantitative credit risk framework to go beyond standard financial and transactional data for their implementation by policy makers and financial institutions. The important contributions of this thesis include theoretical frameworks to evaluate the hardly observable, small firms with higher precision through different data utilization (granular credit event definitions) and data augmentation (collecting and transforming unstructured data). Empirical findings provide actual benefits for multi-region, cross-events evaluation, demonstrating how the focus on robustness can translate into higher confidence outcomes, such as importance of early warning signals, non financial information, and regional effects.

Item Type	Thesis (PhD)
URI	https://centaur.reading.ac.uk/id/eprint/130663
Identification Number/DOI	10.48683/1926.00130663
Divisions	Henley Business School
Date on Title Page	September 2025
Download/View statistics	View download statistics for this item