Correlation and Covariance: How Asset Relationships Drive Diversification

Calculating Correlation and Covariance

Covariance measures how two variables move together: when one is above its average, is the other also above its average? The formula for sample covariance between assets A and B is: sum of (return_A - avg_A) x (return_B - avg_B) / (n-1). A positive covariance means the assets tend to move together. A negative covariance means they tend to move in opposite directions. The magnitude of covariance depends on the units, making it difficult to compare across pairs. Correlation standardizes covariance to a -1 to +1 scale: correlation = covariance / (standard_deviation_A x standard_deviation_B). For example, if the covariance between stocks and bonds is -0.0012, stock volatility is 15%, and bond volatility is 5%, the correlation is -0.0012 / (0.15 x 0.05) = -0.0012 / 0.0075 = -0.16. This means stocks and bonds have a slight negative correlation — a modest but important diversification benefit. The correlation matrix of all assets in a portfolio is the essential input for portfolio optimization. Standard deviation and variance explained →

Correlation Between Major Asset Classes

Understanding typical correlations between asset classes helps investors construct diversified portfolios. US large-cap stocks vs US small-cap stocks: 0.80-0.90 (high — both are equities). US stocks vs international developed stocks: 0.70-0.85 (moderate-high — different economies but same equity risk premium). US stocks vs emerging market stocks: 0.60-0.75 (moderate — more independent economic drivers). US stocks vs long-term Treasury bonds: -0.30 to +0.30 (varies — negative during deflation, positive during inflation). US stocks vs corporate bonds: 0.20-0.50 (moderate — both respond to economic conditions). US stocks vs gold: 0.0-0.10 (near zero — gold is a distinct asset class). US stocks vs commodities: 0.10-0.30 (low — commodities respond to supply/demand cycles). US stocks vs real estate (REITs): 0.50-0.70 (moderate — real estate has equity-like characteristics). US stocks vs cash (T-bills): -0.10 to 0.0 (near zero). These correlations are not stable — they change over time and particularly during market crises. Asset allocation fundamentals →

Correlation During Market Crises

The most important — and dangerous — feature of correlation is that it tends to increase during market crises. This phenomenon is called "correlation convergence" or "correlation to 1." During the 2008 financial crisis, correlations between stock markets worldwide approached 0.9, meaning international diversification provided little benefit when it was needed most. During the 2020 COVID crash, correlations between risk assets (stocks, corporate bonds, real estate, commodities) all converged above 0.8. During 2022, the historically negative stock-bond correlation turned positive as both fell together due to rising interest rates, breaking the traditional 60/40 diversification hedge. This behavior means that the diversification benefit calculated from long-term average correlations overstates the benefit available during crises. Sophisticated investors account for this by using stress-test correlations (the correlations observed during past crises) rather than long-term average correlations, and by including assets that tend to maintain their diversification properties during crises — typically long-term Treasuries, gold, and the US dollar. Managing portfolio risk →

Using Correlation in Portfolio Construction

Correlation is the essential input for portfolio optimization. The goal of mean-variance optimization is to find the portfolio weights that minimize portfolio variance for a given expected return, or maximize expected return for a given variance. The portfolio variance formula for N assets is: sum of (w_i^2 x sigma_i^2) + sum of sum of (2 x w_i x w_j x covariance_ij). This formula shows that adding an asset with low or negative correlation to the existing portfolio reduces overall portfolio variance even if the new asset has higher individual variance. This is the mathematical justification for including alternative assets like gold, managed futures, and trend-following strategies in a portfolio even though they have lower expected returns than stocks. The practical application: investors should prioritize finding assets with low correlation to their existing holdings, not just assets with high expected returns. Correlation determines whether a new investment improves or worsens portfolio efficiency. Risk parity: balancing risk contributions →

What is the difference between correlation and covariance?

Covariance is the raw measure of how two assets move together, but its magnitude depends on the units of measurement (percentage returns). Two volatile assets can have a large covariance even with moderate correlation, while two low-volatility assets can have a small covariance even with high correlation. Correlation standardizes covariance to a -1 to +1 scale by dividing by the product of the standard deviations, making it comparable across different asset pairs. Correlation is the more useful metric for investment decisions because it allows you to compare the relationship between stocks and bonds (correlation -0.1 to +0.3) with the relationship between US and international stocks (correlation 0.7-0.85). Investors should use correlation for asset allocation decisions and use covariance for mathematical portfolio optimization calculations. Both are essential tools for understanding the risk-reducing benefits of diversification.

What does a correlation of 0 mean?

A correlation of 0 means there is no linear relationship between the returns of two assets. When one asset goes up, the other is equally likely to go up, go down, or stay flat. This does not mean the assets are independent in all respects — there could be a non-linear relationship that correlation does not capture. A correlation of 0 provides excellent diversification benefits because the portfolio variance formula's covariance term becomes zero, leaving only the weighted individual variances. Historically, the correlation between US stocks and gold has been near zero (0.0-0.1), which is why gold is valued as a portfolio diversifier. The correlation between US stocks and cash (T-bills) is also near zero. A zero-correlation asset added to a portfolio reduces risk regardless of its own volatility, making it extremely valuable for portfolio construction even if its expected return is modest. Gold as a portfolio diversifier →

Can I rely on historical correlations for future portfolio construction?

Historical correlations provide useful guidance but should not be relied upon mechanistically. Correlations between asset classes are not stable over time — they change with economic regimes, monetary policy, and market conditions. The stock-bond correlation has fluctuated between -0.5 and +0.5 over different decades. Correlations during calm markets differ significantly from correlations during crises. Currency correlations shift with monetary policy divergence. The practical approach is to: use long-term average correlations as a starting point, stress-test with crisis-period correlations (2008, 2020, 2022), use rolling correlations to identify trends, and incorporate judgment about the current economic regime. Many sophisticated investors use regime-dependent correlations — different correlation assumptions for inflationary, deflationary, growth, and recessionary environments. The key lesson: diversification strategies should be robust to a range of correlation assumptions, not optimized for a single historical estimate. Asset allocation by goal and regime →

How many assets do I need to benefit from correlation?

The number of assets needed depends on the average correlation between them. If assets have high average correlation (like US stocks at 0.5-0.6 average pairwise correlation), you need more assets (20-50) to achieve most of the available diversification benefit. If assets have low average correlation (like a multi-asset portfolio blending stocks, bonds, commodities, and alternatives with average pairwise correlation of 0.2-0.3), you can achieve significant diversification with as few as 5-10 assets. The marginal benefit of each additional asset depends on its correlation with the existing portfolio — adding a gold ETF to a stock-bond portfolio provides more benefit than adding another stock fund, because gold has much lower correlation with the portfolio. The goal is not to maximize the number of holdings but to include assets with genuinely different return drivers: different geographies, different asset classes, and different factor exposures. The three-fund portfolio: simplicity and diversification →

Related Resources