Correlation Regimes
When correlations break — and what diversification means then
The cleanest empirical finding in cross-market risk research is also the most uncomfortable one: pair-wise asset correlations are not constant. They vary materially across calm and stressed regimes, and the variation runs in the wrong direction for investors. In calm markets, correlations across asset classes and across stocks are low — the textbook diversification benefit is genuine.
In stressed markets, correlations rise sharply — sometimes from 0.30 to 0.85 within the cross-section of a single asset class, and sometimes across entirely different asset classes that had been treated as orthogonal.
The phenomenon has been called 'correlation breakdown' (when historically-stable correlations shift) and 'correlation contagion' (when stress in one asset class spreads to others through repositioning rather than fundamental connection). Either name captures the structural fact: the diversification a portfolio shows in calm markets is systematically more than the diversification it has during stress, and the difference is often enough to render the calm-market diversification meaningless precisely when an investor needs it most.
That is the opening. Finishing a lesson is where it stops being interesting and starts being useful: the full lesson runs to 8 sections and ends with 6 practice questions. A free account is what opens the rest, and the other 255 lessons in the Academy with it. No card.
What this lesson covers
- 1What drives correlation regime shifts
- 2Cross-asset-class correlations are not exempt
- 3Correlation Regime Visualizer
- 4Detecting and modeling correlation regimes
- 5Pair-wise correlation by regime — illustrative U.S. asset-class behavior
- 6October 2008 — when correlations went to one
- 7Where to see this on the platform
- 8Summary