Conditional correlation is the co-movement between assets measured within a specific market state or regime rather than over the full sample. It captures the empirical fact that correlations are not stable: cross-asset and cross-stock linkages typically rise sharply in stress states, so the relevant number is the one that holds during the drawdown, not the calm-period average.
How it works
Formally, it is the correlation of returns conditioned on an event or regime variable — e.g. correlation given the market is in its lower return tail. Unconditional (full-sample) correlation blends benign and stressed periods and understates joint downside risk; models such as DCC-GARCH estimate how the conditioning correlation evolves through time. The key asymmetry: correlations cluster toward one precisely when diversification is most needed.
Why it matters now
With index ownership concentrated in passive vehicles and the Magnificent Seven dominating cap-weighted benchmarks, the conditional correlation of a future drawdown is higher than placid realised dispersion suggests — mechanical, flow-driven selling raises the odds that names fall together when the tape turns.
Example
In the 2008 crisis, equity pairwise correlations that averaged roughly 0.3–0.4 in calm periods spiked above 0.8 as markets fell, and the same regime shift recurred in March 2020. A portfolio sized on the unconditional 0.35 would have been carrying far more joint downside than that single number implied once the stress state arrived.
Frequently asked
- What is conditional correlation?
- Conditional correlation is the co-movement between assets measured within a specific market regime rather than across the full sample. It captures the fact that correlations are unstable: cross-asset linkages spike toward one during stress states. In 2008 and again in March 2020, equity pairwise correlations that averaged 0.3–0.4 in calm periods jumped above 0.8 as markets fell.
- How does conditional correlation differ from unconditional correlation?
- Conditional correlation measures co-movement within a defined state or event, while unconditional correlation averages across the entire sample. The unconditional number blends calm and stressed periods and understates joint downside risk. A pair averaging 0.35 unconditionally may run above 0.8 in a drawdown, so the conditional estimate is the one that governs tail risk
Glossary · downside tail risk
Downside tail risk is the probability of low-likelihood, high-severity adverse outcomes residing in the left tail of a return or macro distribution. It captures losses far beyond normal variance — the rare but ruinous events that Gaussian models systematically underweight and that dominate portfolio and policy risk.
Read more →
- Why does conditional correlation matter for portfolio risk?
- Conditional correlation matters because diversification fails precisely when it is needed most. Correlations cluster toward one in stress states, so a portfolio sized on placid-period averages carries far more joint downside than that single number implies. Sizing positions on a calm-period 0.35 instead of a stressed 0.8 systematically understates drawdown exposure when the regime shifts.
- How is conditional correlation estimated?
- Conditional correlation is estimated by conditioning returns on a regime or event variable, such as the market being in its lower return tail. Dynamic models like DCC-GARCH (dynamic conditional correlation) let the correlation matrix evolve through time rather than holding it fixed. The key empirical finding is asymmetry: correlations rise faster in down markets than up markets.
- Why is conditional correlation higher now with passive ownership and the Magnificent Seven?
- Conditional correlation is elevated because index concentration and passive flows create mechanical, flow-driven selling that pushes names down together. With the Magnificent Seven dominating cap-weighted benchmarks, a drawdown's conditional correlation exceeds what calm-period realised dispersion suggests. The same baskets that diversify on the way up converge sharply when redemptions force simultaneous selling.