Latin American sovereign stress avalanches follow power-law size distribution with exponent 1.77, synchronized via common factors not direct propagation

The preprint applies finite-sample power-law diagnostics and minimum-spanning-tree networks to country-level stress events defined by log-spread innovations exceeding volatility thresholds. Rolling correlations reveal denser, spectrally amplifying structures during large avalanches, yet partial-correlation filtering eliminates this pattern, pointing to common-factor co-movement. A country-level reshuffling placebo confirms synchronization exceeds independent timing expectations. These results frame sovereign fragility as finite-size criticality amenable to monitoring rather than early-warning prediction.

Mainstream coverage of Latin American debt often emphasizes bilateral contagion channels or individual country fundamentals while under-quantifying avalanche statistics. The observed heavy tails align with documented patterns in equity and FX networks but extend them to EMBI spreads, revealing that large events coincide with elevated raw connectivity without residual propagation after factor removal. This distinguishes the work from cascade models that assume direct spillovers.

The analysis connects to broader emerging-market literature on external shocks and commodity cycles, where common-factor dominance has historically amplified stress during global risk-off episodes. Network metrics here describe contemporaneous regimes rather than leading indicators, a distinction that tempers their use in real-time surveillance. Strengthening evidence would require out-of-sample testing on post-2026 data and explicit factor identification.

Future monitoring should track whether avalanche frequency rises above historical baselines when U.S. rate volatility increases, testing the criticality framework directly.