Predicting Financial Crises in Iraqi Banks Using the ARIMA Model: An Applied Study on the Non-Performing Loans 2011-2024
Ibrahim Baraka Hamadi
*
College of Administration and Economics, University of Misan, Amarah, Iraq.
*Author to whom correspondence should be addressed.
Abstract
The Iraqi banking sector plays a vital role in maintaining financial stability and supporting economic development. However, rising credit risks require reliable forecasting tools capable of strengthening early warning systems. Despite the widespread application of time-series forecasting models, empirical evidence on predicting non-performing loan (NPL) ratios in Iraqi banks remains limited.
This study evaluates the predictive performance of the Autoregressive Integrated Moving Average (ARIMA) model in forecasting NPL ratios and identifies the most appropriate model specification for each bank. Annual data for four Iraqi commercial banks listed on the Iraq Stock Exchange covering the period 2011–2024 were collected from the Central Bank of Iraq, the Iraq Stock Exchange, and the banks’ annual reports.
Stationarity was examined using the Augmented Dickey–Fuller (ADF) test, while ARIMA specifications were selected based on autocorrelation and partial autocorrelation analyses supported by model selection criteria. Forecast accuracy was evaluated using the coefficient of determination (R²), root mean square error (RMSE), mean absolute error (MAE), mean absolute percentage error (MAPE), and Theil’s U² coefficient, followed by three-year forecasts for each bank.
The estimated ARIMA models produced statistically acceptable forecasting performance. Gulf Commercial Bank achieved the highest explanatory power (R² = 0.818), whereas Iraqi Investment Bank recorded the lowest forecasting error (MAPE = 33.61%; Theil’s U² = 0.970). Forecasts indicate a gradual increase in NPL ratios during 2025–2027, suggesting a deterioration in loan portfolio quality. These findings demonstrate that NPL ratios can serve as effective quantitative early warning indicators of increasing credit risk rather than direct measures of financial crises. Accordingly, ARIMA models provide a practical and transparent forecasting framework for supporting banking supervision and credit risk monitoring in Iraqi commercial banks.
Keywords: ARIMA, non-performing loans, credit risk, financial forecasting, Iraqi commercial banks, early warning systems, time-series analysis, banking supervision, forecast accuracy, financial stability