Examples and Applications of Correlation
You'll learn how to apply correlation concepts to solve real-world problems involving portfolio variance, bivariate normal distributions, and contextual interpretation.
Core Idea
Now that you understand covariance and correlation coefficient properties, it's time to see them in action. Three key applications demonstrate correlation's power: managing financial risk through portfolio variance, working with bivariate normal distributions (where correlation fully characterizes joint behavior), and interpreting correlation meaningfully in context—knowing when it's strong, what it implies, and what it doesn't imply.
Portfolio Variance
When combining two investments with returns X and Y into a portfolio (investing fraction w in X and 1-w in Y), the portfolio return is R = wX + (1-w)Y. Using the variance of sums formula:
Var(R) = w²Var(X) + (1-w)²Var(Y) + 2w(1-w)Cov(X,Y)
The correlation ρ = Cov(X,Y)/(σ_X σ_Y) determines diversification benefit. When ρ < 1, combining assets reduces total risk—negative correlation provides the best diversification!
Bivariate Normal Distributions
For jointly normal random variables (X,Y), the correlation coefficient ρ completely determines their dependence structure beyond the marginals. The joint density involves ρ explicitly, and conditional distributions are normal with means that shift linearly: E[Y|X=x] moves proportionally to ρ.
Interpreting Correlation in Context
- Magnitude: |ρ| near 1 indicates strong linear association; near 0 suggests weak linear relationship
- Not causation: High correlation doesn't imply X causes Y
- Linear only: ρ = 0 means no linear relationship, but nonlinear dependence may exist
- Context matters: ρ = 0.3 might be significant in physics but weak in psychology
Key Takeaway: Correlation quantifies linear association and is essential for portfolio optimization, characterizes bivariate normal dependence, but must be interpreted carefully—it measures only linear relationships and never implies causation.