Designing a Financial Stress Index Based on the GHARCH-DCC Approach and Machine Learning Models

This research focuses on designing a financial stress index using the GHARCH-DCC approach and machine learning models to predict financial crises. This study creates a composite index to measure the Iranian financial system and its turbulent effects in uncertain conditions of the Tehran Stock Exchange during the years 2009 to 2020. Various turbulent factors, including exchange rates, stock indices, the banking industry, gold prices, energy carriers, and the insurance industry, are used as variables. By combining the GHARCH-DCC approach with the ANN approach, the best predictive model for the financial stress index is created. The results indicate a significant and positive impact of all independent variables except for gold price turbulence on the stress index. The model’s coefficient of determination indicates a good fit. The findings demonstrate significant periods of financial stress, with the highest stress occurring in 2018. From 2018 to 2020, a considerable increase in stress compared to recent years has been observed. This research provides valuable insights into financial stress and helps assess risks and make policy decisions.