Correlation risk

Correlation risk is the risk that the statistical relationships between assets or institutions will change in ways that undermine diversification and amplify systemic losses. In normal market conditions, correlations between asset classes are low or even negative, allowing portfolio diversification to reduce overall risk. During crises, correlations tend to converge toward one — every asset falls simultaneously, every counterparty weakens together — and the diversification benefit that was priced into the system vanishes precisely when it is most needed.

This dynamic is not merely a statistical curiosity; it is a structural feature of networked financial systems. When institutions are connected through credit default swap contracts, collateralized debt obligations, and interbank lending, the default of one node triggers forced asset sales by its counterparties, which depresses prices, which increases the leverage of other nodes, which triggers further sales. The network topology creates endogenous correlation: the connections themselves generate the covariance that statistical models treat as exogenous. The result is that value-at-risk models, which rely on historical correlation matrices, systematically underestimate tail risk because they assume the future will resemble the past — an assumption that collapses when the network structure itself becomes the source of correlation.

Correlation risk is the most dangerous risk because it is the risk that your risk model does not know it has. Every portfolio optimized against historical correlations carries a hidden exposure to network topology — and that exposure only reveals itself when the network is already on fire.