Standard deviation measures how much investment returns swing around their average — the higher the number, the wider the range of possible outcomes. For retirement planning, volatility matters because it creates variance drain and amplifies sequence-of-returns risk, both of which can derail a withdrawal plan.
Standard deviation is the most common statistical measure of investment volatility — how much returns deviate from their average over time. A stock portfolio with a 10% expected return and 15% standard deviation will see annual returns between -5% and +25% roughly two-thirds of the time (one standard deviation). The remaining one-third of years fall outside this range, sometimes far outside.
How It Works
Standard deviation is calculated as the square root of the variance of returns. In practical terms:
- 68% of returns fall within ±1 standard deviation of the mean
- 95% of returns fall within ±2 standard deviations
- 99.7% of returns fall within ±3 standard deviations (under a normal distribution)
| Asset Class | Expected Return | Standard Deviation | Typical Annual Range (68%) |
|---|---|---|---|
| Large-cap stocks | 10% | 16% | -6% to +26% |
| Bonds | 4% | 6% | -2% to +10% |
| Cash | 2% | 1% | +1% to +3% |
| 60/40 portfolio | 7.6% | 10% | -2.4% to +17.6% |
Higher standard deviation means a wider range of possible outcomes in any given year. For accumulators with decades ahead, this volatility averages out. For retirees withdrawing from the portfolio, volatility is a much bigger problem.
Why It Matters for Retirement Planning
Volatility affects retirees in two critical ways:
-
Variance drain: Higher volatility reduces compound (geometric) returns even if the arithmetic average stays the same. A portfolio alternating between +30% and -10% has a 10% arithmetic average but only 8.2% geometric growth. The difference is approximately half the variance.
-
Sequence-of-returns risk: When withdrawals combine with volatile returns, the order of returns matters enormously. Poor returns early in retirement — which are more likely with higher volatility — can permanently impair the portfolio.
This is why asset allocation decisions involve balancing the higher expected returns of equities against their higher volatility. A 100% stock portfolio has the highest expected return but also the highest drawdown risk during the critical early retirement years.
A Practical Example
Two portfolios both target 8% average returns but with different volatility:
| Year | Low Volatility (8% SD) | High Volatility (20% SD) |
|---|---|---|
| 1 | +12% | +30% |
| 2 | +4% | -15% |
| 3 | +10% | +25% |
| 4 | +6% | -5% |
| Arithmetic avg | 8% | 8.75% |
| Geometric avg | 7.97% | 7.65% |
Despite a higher arithmetic average, the high-volatility portfolio compounds at a lower rate. For a retiree withdrawing $40,000/year from a $1,000,000 portfolio, the high-volatility path produces a worse outcome in the majority of simulated scenarios due to the combination of variance drain and withdrawal timing.
Frequently Asked Questions
- What is a typical standard deviation for a stock portfolio?
- U.S. large-cap stocks have a historical annual standard deviation of roughly 15-17%. Bonds are around 5-7%. A balanced 60/40 stock/bond portfolio typically has a standard deviation of 9-11%. International and small-cap stocks tend to be higher at 18-22%.
- Is higher volatility always bad for retirement portfolios?
- Not necessarily — volatility is the price you pay for higher expected returns. Stocks are more volatile than bonds but historically deliver much higher long-term returns. The problem is when high volatility combines with portfolio withdrawals, creating sequence-of-returns risk. The key is matching volatility to your time horizon and withdrawal needs.
- How does standard deviation relate to fat tails?
- Standard deviation measures the average spread of returns, but it doesn't capture how frequently extreme events occur. Two distributions can have the same standard deviation but very different tail behavior. Fat-tail distributions (high kurtosis) produce more extreme events than the standard deviation alone would suggest.