Cyberattacks as Systemic Shocks: An AI-Generated Cyber-Risk Index and Confidence-Contagion Model in Semi-Peripheral Financial Systems

Cyberattacks are increasingly framed as systemic financial-stability risks, yet empirical work linking cyber-incident discourse to macro-financial contagion remains confined to single-country case studies, firm-level disclosure regressions, or network models calibrated to advanced economies. We address this gap with a Cyber Risk Index (CRI) for eight semi-peripheral economies across Latin America and Central-Eastern Europe (2015–2025), derived from GDELT news discourse via a validated two-stage retrieval-and-classification pipeline. The CRI is embedded in an explicit causal-loop structure linking discourse intensity to confidence erosion, liquidity stress, market Value-at-Risk, exchange-rate pass-through, and monetary-policy response, estimated via a fixed-effects panel with Driscoll-Kraay errors (777 country-month observations, near-complete 2015–2025 coverage). The panel shows substantial cross-country heterogeneity — Romania highest in cyber-discourse intensity, Peru highest in market and exchange-rate volatility, an inverse pattern warranting further study. A contemporaneous specification finds no significant CRI relationship with equity Value-at-Risk and only a marginal, sign-unstable relationship with exchange-rate volatility that vanishes under one- to three-month lags and remains stable as the sample nearly doubled — reported as a null result rather than reframed as positive. The study contributes a validated method for AI-generated cyber-risk indicators in data-scarce contexts and the first cross-country systemic-risk evidence for the semi-periphery, alongside a finding that motivates richer specifications of the proposed causal loop.