Aug 2026· International Journal of Science and Research Archive· 0 citations
TL;DR
The paper discusses a dynamic stress test using Generative Adversarial Networks (GANs), answers key questions in the context of Explainable AI (XAI) and data privacy, and offers a recommended course of action in implementing and standardised RegTech/SupTech systems to transition regulatory control to an active and proactive science.
Abstract
The world financial environment is characterised by increasing interdependence, rapid technological change and cyclic economic crises such that current and legacy EWS is no longer useful in regard to macroeconomic governance. Conventional models, based on lagging indicators and historical data, do not take into account contagion non-linearities, velocity and systemic shocks in the digital world, such as algorithmic “flash” crashes or “bank” runs. The paper suggests the use of AI to overcome these issues, through real-time monitoring and early warning of risks. The suggested Multi-layered structure to combine Big Data, Machine Learning and Artificial Intelligence in the macroeconomic direction is based on the Financial Instability Hypothesis (FIH) of Minsky and the theory of Complex Adaptive Systems (CAS). The architecture comprises data ingestion, core analytics (unsupervised anomaly detection, network contagion mapping and supervised crisis forecasting) and decisions support operational dashboards. Moreover, it discusses a dynamic stress test using Generative Adversarial Networks (GANs), answers key questions in the context of Explainable AI (XAI) and data privacy, and offers a recommended course of action in implementing and standardised RegTech/SupTech systems to transition regulatory control to an active and proactive science.
Stress testing is a critical component of banking risk management and supervisory oversight, providing a structured means of evaluating whether financial institutions can absorb losses and maintain capital resilience under adverse conditions. Yet the increasing speed, complexity, and interconnectedness of financial markets raise questions about the ability of predominantly scenario-based and periodically executed approaches to identify emerging vulnerabilities. This conceptual paper examines how artificial intelligence (AI), machine learning (ML), alternative data, and advanced analytics could augment conventional bank stress testing by enabling more adaptive scenario generation, nonlinear risk estimation, continuous monitoring, and integrated portfolio analysis. The paper uses a structured conceptual synthesis of the stress-testing literature, supervisory principles, and recent research on AI-enabled financial risk modeling to identify limitations in traditional approaches and develop a framework for AI-enabled dynamic stress testing. The proposed framework links data and risk sensing, AI-assisted scenario generation, portfolio and balance-sheet simulation, capital and liquidity impact assessment, management decision-making, and model-risk governance. The analysis argues that AI should complement rather than replace established stress-testing governance and supervisory processes. In particular, model validation, explainability, data quality, human oversight, documentation, and ongoing monitoring are essential for responsible adoption. The paper contributes a research-oriented framework for understanding how banks can move from predominantly periodic stress exercises toward more adaptive and forward-looking risk intelligence while preserving regulatory discipline. It also identifies empirical research opportunities, including comparative testing of AI and traditional models, evaluation of scenario-generation quality, and measurement of governance effectiveness.
Nishitha Kambalapally· Journal of Economics, Financ...· 0 citations
Banking crises are a persistent threat to macroeconomic stability in emerging markets, where conventional econometric monitoring frameworks often fail to capture non-linear macro-financial relationships. This paper examines whether machine learning algorithms can improve the detection of banking crisis risk in Nigeria compared to standard logistic regression. We compare the performance of Random Forest, Support Vector Machine (SVM), and Extreme Gradient Boosting (XGBoost) against logistic regression using annual data from the African Financial Crises dataset (1954–2014). Resampling is only implemented on the training set to overcome the infrequency of crisis events. Performance on models is assessed based on accuracy, precision, recall, F1-score, and the area under the receiver operating characteristic curve (AUC) in a rigorous out-of-time validation setting. Our findings indicate that tree-based ensemble models outperform logistic regression on the test set: XGBoost achieves the best generalization performance (AUC = 1.0; F1 = 0.95 in non-crisis, 0.80 in crisis), whereas Random Forest has the highest cross-validated F1-score on the training set. The most important variables are exchange rate volatility, inflation, and indicators of systemic crisis. The most significant crisis indicators are, however, seen in crisis years, which means that the annual data do not provide much lead-time to detect the crisis. These results should be taken with caution because of the small sample size and the limited number of crisis observations during the test period. Altogether, machine learning models have potential as additional tools to monitor banking crises in Nigeria, though at the moment they are not fully operational as policy instruments.
Ntanganedzeni Mandiwana, Thakhani Ravele, C. Sigauke et al.· Analytics· 0 citations
Accurately predicting economic growth and unemployment rate is an important prerequisite for macroeconomic regulation in financial social networks. Traditional security monitoring and warning methods often rely on lagging official statistical data, making it difficult to capture nonlinear correlations and sudden signals in economic dynamics in real time. Therefore, this paper proposes a Boruta-SHAP and Transformer-XL (BST-XL) security monitoring and warning model based on Transformer-XL. The framework first constructs a Boruta-SHAP (BS) two-stage feature selection method based on ensemble learning framework and Shapley Additive Explanations (SHAP) interpretability analysis. By identifying important features related to economic time series task from a given feature set, focusing on economic growth and unemployment rate, the problem of feature redundancy in economic time series data is effectively solved. Secondly, by using Transformer-XL to measure the impact of historical economic sequences on recent economic dynamics, the latest state characteristics of economic dynamics can be obtained. This is beneficial for accurately predicting the dynamic changes of economic growth and unemployment rates. Experimental analysis shows that BST-XL performs well in predicting economic growth and unemployment rate, with higher security monitoring accuracy. It is suitable as a predictive model for economic growth and unemployment rate in financial social networks.
Lei Liu, Yuanyuan Wen· Frontiers of Physics· 0 citations
Macroeconomic forecasting has become increasingly challenging in environments characterized by structural breaks, nonlinear dynamics, and elevated economic uncertainty. Traditional econometric forecasting models frequently experience substantial predictive deterioration during periods of financial crises, geopolitical instability, and rapidly evolving macroeconomic conditions due to their reliance on assumptions of parameter stability and linear economic relationships. In response to these limitations, this study proposes an integrated causal machine learning framework designed to improve macroeconomic forecasting performance under structural instability and uncertainty. The proposed framework combines structural break detection techniques, machine learning algorithms, causal inference methodologies, and Explainable Artificial Intelligence (XAI) tools within a unified empirical architecture. More specifically, the study integrates Bai–Perron structural break analysis, Markov-Switching regime identification, Double Machine Learning (DML), Causal Forest estimation procedures, and SHAP-based explainability techniques. The empirical analysis employs a U.S. macroeconomic time-series dataset covering major crisis episodes, including the 2008 Global Financial Crisis, the COVID-19 pandemic, and the 2022 inflation shock. The dataset combines inflation, monetary, financial, energy-market, and uncertainty indicators obtained from publicly available U.S. macroeconomic databases. The empirical findings demonstrate that causal machine learning models significantly outperform conventional econometric frameworks such as VAR and TVP-VAR models, as well as standard machine learning algorithms including Random Forest (RF), XGBoost, and LSTM networks. The Double Machine Learning framework generates the strongest forecasting performance across all forecasting horizons, economic regimes, and robustness specifications. The results further reveal that macroeconomic relationships are highly regime-dependent and strongly influenced by uncertainty indicators, financial volatility, oil price shocks, and monetary policy dynamics. Explainability analysis additionally shows that uncertainty measures and energy market variables become dominant drivers of inflation forecasts during crisis periods characterized by elevated instability. The study contributes to the growing literature on macroeconomic forecasting by bridging econometric forecasting theory, causal inference methodologies, machine learning techniques, and explainable artificial intelligence within a unified forecasting framework. The findings provide important implications for central banks, policymakers, and financial institutions seeking more adaptive, transparent, and robust forecasting systems under uncertain macroeconomic environments.
Financial crises have a significant impact on economic stability, making early detection crucial for policymakers and financial institutions. This study aims to develop a Financial Crisis EarlyWarning System by integrating time-varying dependency structures with machine learning techniques for proactive systemic risk detection. This study uses monthly macro-financial indicators from January 2020 to May 2025. The methodology includes rolling copula estimation to capture dynamic tail dependencies, feature engineering to transform copula parameters into systemic risk indicators, and predictive modeling using Backpropagation Neural Networks and Long Short-Term Memory. To ensure robust validation, the dataset is split chronologically (70% training, 30% out-of-sample testing). The results show that Frank Copula optimally captures nonlinear tail dependencies, particularly between stock indices and exchange rates. The predictive model achieves high out-of-sample detection accuracy (Backpropagation Neural Network: 91.2%; Long Short-Term Memory: 92.5%). Backpropagation Neural Networks successfully identified acute short-term shocks during validated crisis episodes, while Long Short- Term Memory detected long-term structural vulnerabilities and projected severe systemic stress in forward-looking forecasts. Conclusion: Integrating copula-based feature engineering with a sequence learning architecture significantly improves early warning accuracy. This integrated framework provides policymakers, regulators, and market participants with a powerful and proactive tool for financial stability surveillance and crisis mitigation.
In the context of increasing global economic uncertainty, the financial risks of listed companies exhibit significant dynamic evolution characteristics. Traditional methods based on static indicators are no longer sufficient to meet the requirements of forward-looking early-warning. Considering the intertemporal transmission and path dependence characteristics of enterprise financial risks, from the perspective of deep temporal modeling, this study constructs a Gated Recurrent Unit model (Residual GRU, RGRU) that integrates a residual connection mechanism to enhance the feature transmission ability in the deep network and alleviate the problem of gradient propagation difficulties. Taking Chinese A-share listed companies from the first quarter of 2010 to the fourth quarter of 2024 as the research sample, and using the Special Treatment (ST) or Delisting Risk Warning (*ST) status as the distress determination criterion, an enterprise financial distress early-warning system is constructed based on multi-dimensional financial and corporate governance indicators. On this basis, ablation experiments are set up from two dimensions—the type of recurrent units (GRU and LSTM) and the connection mechanism (residual connection and dense connection)—and performance is evaluated through systematic parameter optimization and multi-layer network structure experiments. The empirical results show that the RGRU model outperforms the comparison model in classification indicators such as accuracy, F1 score, recall, and probability prediction indicators such as Brillouin score and logarithmic loss, and demonstrates better stability and generalizability. The research results indicate that the introduction of residual connections can effectively improve the prediction performance and practical value of recurrent neural networks in the enterprise financial distress early-warning task.
A new method, called CW-Net, translates the reasoning process of an autonomous vehicle’s AI system into understandable concepts that explain its behavior.