ABSTRACT
The threats posed by financial crises to the stability of economies have necessitated their early detection and prediction as important goals for both policymakers and financial institutions. In this regard, this study bases its analysis on the dependent variable FSI as the index most comprehensive in its measurement of systemic risk. Other key economic indicators used as independent variables include GDP growth, unemployment, inflation, consumer confidence, and adjusted net national income per capita to understand their impact on financial stress levels. The traditional econometric methods include linear and logistic regression, which are augmented with more complex machine learning models, like the Random Forest and Gradient Boosting, to improve predictive accuracy and interpretation. This study thus highlights the need for an integration of macroeconomic, behavioral, and financial indicators toward capturing the multifaceted nature of financial stress. The inclusion of lagged variables and time-series techniques addresses the temporal dynamics usually overlooked in earlier studies. The findings show that the Consumer Confidence Index is the most significant predictor of a financial crisis, followed by inflation and income growth, indicating that behavioral and monetary factors are crucial for an accurate prediction.
