Correlation
Correlation is a statistical measure of the linear relationship between two random variables, normalized to fall between −1 (perfectly negatively correlated) and +1 (perfectly positively correlated), with 0 indicating no linear relationship. In finance, correlation is the fundamental input to portfolio diversification theory, joint risk modeling, derivatives pricing, and stress testing.
Key takeaways
- Pearson's correlation coefficient ρ = Cov(X,Y) / (σ_X × σ_Y) measures linear co-movement; it is the most common measure in finance but may miss nonlinear dependence.
- Correlation is not causation — two assets may be correlated due to common exposure to an underlying factor rather than any direct relationship.
- Portfolio variance = w₁²σ₁² + w₂²σ₂² + 2w₁w₂ρσ₁σ₂; the benefit of diversification grows as ρ decreases toward −1.
- Correlations are notoriously unstable: they increase sharply during market crises (contagion), precisely when diversification is most needed.
- Implied correlation (derived from dispersion options or correlation swaps) reflects the market's forward-looking view of correlation and is tradeable.
Explanation
Pearson's correlation coefficient is defined as:
ρ_{X,Y} = Cov(X,Y) / (σ_X × σ_Y) = E[(X−μ_X)(Y−μ_Y)] / (σ_X × σ_Y)
In empirical applications, the sample correlation is computed from historical return data. For n observations of returns x_i and y_i:
ρ = [Σ(x_i − x̄)(y_i − ȳ)] / [√(Σ(x_i − x̄)²) × √(Σ(y_i − ȳ)²)]
The diversification implication is the central insight of Modern Portfolio Theory. For a two-asset portfolio with equal weights (50/50) and equal volatilities σ, portfolio volatility = σ × √((1 + ρ)/2). At ρ = 1 (perfect positive correlation), portfolio volatility = σ (no diversification benefit). At ρ = 0 (no correlation), portfolio volatility = σ/√2 ≈ 0.707σ (approximately 30% reduction). At ρ = −1, portfolio volatility = 0 (perfect hedge). This shows that most diversification benefit is captured at moderate negative correlations — very negative correlations are rare and unstable.
Correlation instability is a major practical challenge. Correlations estimated from 3-year historical windows represent a blended average that may be dominated by specific market regimes. During the 2008 financial crisis, equity correlations within developed market indices rose from historical averages of 0.3–0.4 to above 0.8 as forced selling caused simultaneous declines across previously uncorrelated sectors. Hedge fund long/short books that relied on 0.3 pairwise correlation for risk calculation found their actual portfolio volatility 50–80% higher than predicted.
Beyond linear correlation, practitioners use Spearman's rank correlation (which captures monotonic nonlinear relationships), Kendall's tau (rank concordance measure), and copula-based dependence measures (which capture joint tail behavior) for more robust dependence modeling. The 'correlation breakdown' phenomenon and copula tail dependence are therefore critical considerations in any serious multi-asset risk model.
Formula
ρ_{X,Y} = Cov(X,Y) / (σ_X × σ_Y) | Portfolio Variance: σ²_p = w₁²σ₁² + w₂²σ₂² + 2w₁w₂ρσ₁σ₂
Example
A risk manager runs a multi-strategy hedge fund with two sub-portfolios: Long/Short Equity ($500M, vol 12% annually) and Global Macro ($300M, vol 15% annually). Historical correlation between the two strategies is ρ = 0.15. Combined portfolio variance = (500)² × (0.12)² + (300)² × (0.15)² + 2 × (500) × (300) × 0.15 × (0.12) × (0.15) = 3,600 + 2,025 + 810 = 6,435 (in squared $ millions times vol²). Portfolio volatility = √(6,435) × (1/800 of NAV scaling) = approximately 10.0% annualized — well below the weighted average of the two strategy vols (500/800 × 12% + 300/800 × 15% = 7.5% + 5.625% = 13.1%), reflecting significant diversification. During a March 2020-style stress event, the correlation spikes to ρ = 0.70, causing portfolio volatility to rise to approximately 12.5% — a 25% underestimate from the 10.0% pre-crisis figure.
Related terms
Black Swan Event Breakdown Conditional Value At Risk Copula Diversification Equity Expected Shortfall Financial Crisis Global Macro Hedge Fund Liquidity Risk Modern Portfolio Theory