Portfolio Variance and the Role of Correlation
In this chapter
Intermediate
Portfolio variance measures the spread of the combined return around its expected value, in squared return units. It depends on how much you put into each asset, how much each asset moves, and whether they tend to move together. Two investments with sizable individual swings can give you a steadier portfolio if some of those swings cancel. The Diversification chapter shows the intuition with shared and offsetting shocks. Here you’ll calculate the spread.
Start with two assets and a common horizon
Use a fixed opening allocation for one period. Suppose all inputs describe hypothetical one-year nominal USD simple total returns including income, before fees and taxes. Assume both positions are held through the period, no outside cash flows, no borrowing or shorting and weights summing to one.
\[ \sigma_p^2=w_A^2\sigma_A^2+w_B^2\sigma_B^2+2w_Aw_B\sigma_{AB}, \qquad \sigma_{AB}=\rho_{AB}\sigma_A\sigma_B. \]
\(w_A,w_B\) are the opening decimal weights. \(\sigma_A,\sigma_B\) are the assets’ standard deviations over the period. \(\sigma_{AB}\) is their covariance and \(\rho_{AB}\) is the correlation between their returns, from −1 to +1 when both have positive spread. \(\sigma_p^2\) is the portfolio’s variance. Its square root, \(\sigma_p\), is the portfolio’s volatility.
To see where the terms come from, imagine both holdings’ deviations from their own means on the same date. The portfolio deviation is the sum of the two weighted deviations. Squaring that sum gives two squared parts and twice the cross-product. Averaging those terms under the joint return distribution gives the variance above. You can work in decimal returns or in percentage points. Here the standard deviations are in points and all three variance terms are in points squared (pp²). The same covariance inputs can be used to measure each asset’s contribution to portfolio volatility.
Worked case. Put the covariance term in the calculation
Put 60% in a stock fund with volatility of 15 percentage points (pp) and 40% in a bond fund with volatility of 10 pp. Assume correlation is +0.25. Covariance is \(0.25(15)(10)=37.50\) pp².
| Contribution | Calculation | Variance contribution (pp²) |
|---|---|---|
| Stock fund | \((0.60)^2(15)^2\) | 81.00 |
| Bond fund | \((0.40)^2(10)^2\) | 16.00 |
| Co-movement | \(2(0.60)(0.40)(37.50)\) | +18.00 |
| Portfolio | Add the three terms | 115.00 |
Taking the root gives \(\sqrt{115}\), or 10.72 pp of annual portfolio volatility. The simple weighted average of 15 and 10 is 13 pp. That value would fit perfect positive correlation for these nonnegative weights. At +0.25, some of the movements can offset.
Isolate what correlation does
Worked case. Keep everything except correlation fixed
The two weighted volatilities are \(0.60(15)=9\) pp and \(0.40(10)=4\) pp. The first two variance terms total 97 pp² in every row below. The last term is \(2(9)(4)\rho_{AB}=72\rho_{AB}\) pp².
| Correlation | Co-movement term (pp²) | Portfolio variance (pp²) | Portfolio volatility (pp) |
|---|---|---|---|
| −1.00 | −72 | 25 | 5.00 |
| −0.50 | −36 | 61 | 7.81 |
| 0.00 | 0 | 97 | 9.85 |
| +0.25 | +18 | 115 | 10.72 |
| +1.00 | +72 | 169 | 13.00 |
At −0.50 the portfolio volatility is lower than the volatility of either fund. At −1.00, the two weighted spreads are still unequal. The exact offset leaves \(9-4=5\) pp of volatility. A zero result at −1 would require equal weighted spreads as well. At zero correlation the cross term vanishes, and the two individual terms still contribute. If you work in decimals instead, the baseline 115 pp² becomes \(115/10{,}000=0.0115\). Its square root is approximately 0.1072, the same 10.72 pp of volatility. Keep covariance and both variance terms on the same scale.
This table isolates one input. Real periods can change both individual volatilities and co-movement. Check that your inputs come from comparable intervals and correspond to the joint scenario you intend to model. Two estimates based on different dates can describe different market conditions.
Practice questions
Choose one answer per question. Round only the final result where needed.
1. Calculate variance and then volatility
Two equally weighted assets each have a volatility of 10 pp and correlation of +0.50. What is portfolio volatility?
2. Keep the negative covariance sign
Weights are 50/50, each own variance is 64 pp² and covariance is −16 pp². What is portfolio variance?
3. Find an exact offset
Asset A has a volatility of 12 pp and asset B has 8 pp. Which weight in A makes volatility zero when correlation is −1?
4. Change only correlation
Two assets with positive volatilities each have a positive portfolio weight. What happens to portfolio variance as correlation rises?
5. Read a zero covariance
You have two equally weighted assets, each with positive variance. Their covariance is zero. Which term remains?
6. Calculate with decimal inputs
A portfolio’s variance is 0.0144 in decimal return units squared. What is its volatility?