Conditional Adversarial Fragility in Financial Machine Learning under Macroeconomic Stress
This addresses the risk of missed high-risk cases in financial decision systems during adverse economic conditions, though it is incremental as it builds on existing adversarial robustness frameworks by adding regime-aware evaluation.
The paper tackles the problem of evaluating adversarial robustness in financial machine learning under nonstationary economic conditions, finding that adversarial vulnerability is systematically amplified during periods of macroeconomic stress, with models showing substantially greater degradation in predictive accuracy and risk-sensitive outcomes under stress regimes.
Machine learning models used in financial decision systems operate in nonstationary economic environments, yet adversarial robustness is typically evaluated under static assumptions. This work introduces Conditional Adversarial Fragility, a regime dependent phenomenon in which adversarial vulnerability is systematically amplified during periods of macroeconomic stress. We propose a regime aware evaluation framework for time indexed tabular financial classification tasks that conditions robustness assessment on external indicators of economic stress. Using volatility based regime segmentation as a proxy for macroeconomic conditions, we evaluate model behavior across calm and stress periods while holding model architecture, attack methodology, and evaluation protocols constant. Baseline predictive performance remains comparable across regimes, indicating that economic stress alone does not induce inherent performance degradation. Under adversarial perturbations, however, models operating during stress regimes exhibit substantially greater degradation across predictive accuracy, operational decision thresholds, and risk sensitive outcomes. We further demonstrate that this amplification propagates to increased false negative rates, elevating the risk of missed high risk cases during adverse conditions. To complement numerical robustness metrics, we introduce an interpretive governance layer based on semantic auditing of model explanations using large language models. Together, these results demonstrate that adversarial robustness in financial machine learning is a regime dependent property and motivate stress aware approaches to model risk assessment in high stakes financial deployments.