Modeling and forecasting of non-performing loans indicators in the risk management system of the banking sector

Abstract

To develop and test a scientific-methodological approach to the analysis, mathematical smoothing and medium-term forecasting of NPL dynamics in the Ukrainian banking system under conditions of economic instability and wartime uncertainty. The study aims to identify long-term trends in credit portfolio quality and assess future changes in the level of non-performing loans.. Design/methodology/approach. The research is based on monthly data of the National Bank of Ukraine for January 2006 – February 2026 across five banking groups. Moving average and exponential smoothing methods were applied to eliminate short-term fluctuations and identify stable trends. Comparative and structural analysis methods were also used to evaluate changes in NPL indicators during different economic periods. Findings. The study identifies the main macroeconomic and regulatory determinants influencing NPL dynamics. Structural breaks caused by changes in the NBU methodology and the nationalisation of PrivatBank were detected. The results indicate a gradual stabilisation of the banking sector despite wartime risks. According to the forecast, the system-wide NPL ratio may stabilise at 22.4% by February 2027. Research limitations. The model is based on simple exponential smoothing and does not account for non-linear relationships, external shocks or additional macroeconomic variables that may affect borrower solvency and credit risk dynamics. Originality/value. The paper proposes an adapted approach to NPL analysis under wartime conditions in Ukraine. The methodology may be used for internal bank risk management, monitoring credit portfolio quality and macroprudential supervision.

JEL Classification: G21, G28, C22, E44.