Artificial Intelligence Integration in Assessment of Financial Stability of Egyptian Banks

The purpose of this study is to evaluate the stability of Egyptian banks over the period from 2010 to 2024, using Z-score as the dependent variable for financial stability, while the key independent variables are Liquidity Ratio (LR), Return on Assets (ROA), Return on Equity (ROE), Non-Performing Loans (NPL), and Capital Adequacy Ratio (CAR). Neural Networks and XGBoost were used as two AI models to measure the importance of those indicators/predictors regarding the stability of Egyptian banks.
Results show that whilst both models appropriately note the importance of profitability, risk management, capital adequacy, and liquidity to bank performance, they do differ in terms of the ranking of these factors. In Neural Network model, the most important feature is Return on Equity_  ROE (0.4521), then comes the Capital Adequacy Ratio_ CAR (0.3435) and in case of XGBoost, The Capital Adequacy Ratio_CAR has the highest importance (0.3500), then comes ROE (0.1900) and LR (0.2700).The two models confirm the lower significance of NPL compared to others, where XGBoost gave the lowest weight (0.0700) and Neural Networks (0.0982). The different feature rank of independent variables was a result of individual models’ sensitivity to dataset and the processing through its unique logarithms which indicates that AI models can help in the assessments of banks’ stability via examining the predictor power of internal factors which represent Profitability, Risk Management, Liquidity Management and Banks’ Capital Adequacy in an efficient manner.