Abstract
A profitable banking industry is essential for a stable financial system. This study examines bank profitability from two perspectives based on the following research question: How are various factors affecting the profitability of Norwegian banks, and which forecasting model predicts profitability with the highest accuracy? An industry analysis and a case study are conducted to address this question.
The industry analysis utilizes panel data from Norwegian banks between 2015 and 2023, highlighting key factors influencing profitability. A factor framework was applied, incorporating bank-specific and macroeconomic factors within a fixed-effect regression model, and was selected to control for observed heterogeneity across banks. The results indicate that cost efficiency, broader scope, increased policy rate, and economic activity positively affect bank profitability. Conversely, increased size and CET1 ratio show a negative impact. The data was divided into categories to explore the effects across different bank sizes, revealing significant variations in how size and macroeconomic factors affect profitability. These identified variations between the size categories create opportunities for further research.
Additionally, the case study uses forecasting models, including ARIMA, ETS, Random Forest (RF), and LSTM, to predict future ROE for three Norwegian banks (DNB, Sparebank 1, and Jæren Sparebank) over the 2025-2028 period. The models demonstrate varying accuracy levels, with ARIMA and ETS providing stable projections, RF predicting moderate growth, and LSTM showing high volatility. These results suggest that forecasting accuracy depends heavily on the chosen model, with machine learning models such as RF offering more nuanced predictions but with greater variability. The findings from this study contribute to a deeper understanding of the factors that drive bank profitability and guidance for selecting forecasting models for financial analysis.
