Kurtosis measures how often extreme market events occur compared to a normal bell curve. Financial markets have high kurtosis - crashes and rallies happen far more frequently than standard models predict. Ignoring this in retirement simulations can make a plan look safer than it is.
Kurtosis measures the "tailedness" of a probability distribution - specifically, how much of the distribution's variance is driven by extreme outliers versus moderate deviations near the center. In retirement planning, kurtosis quantifies how often market crashes and rallies occur relative to what a normal (bell-curve) distribution would predict.
How It Works
Mathematically, kurtosis is the fourth standardized moment of a distribution - it takes each data point's deviation from the mean, raises it to the fourth power, and normalizes by the variance squared:
A normal distribution has a kurtosis of exactly 3. To make comparisons easier, we often use excess kurtosis (kurtosis minus 3):
| Distribution | Excess Kurtosis | Tail Behavior |
|---|---|---|
| Normal (Gaussian) | 0 | Baseline - thin tails |
| Student-t (DOF=5) | 6 | Fat tails - extreme events ~4x more frequent |
| Student-t (DOF=4) | ∞ (undefined) | Very fat tails |
| Financial returns (empirical) | 3-10 | Significantly fatter than normal |
- Excess kurtosis > 0 (leptokurtic): fatter tails, more extreme events than a normal distribution
- Excess kurtosis = 0 (mesokurtic): normal distribution
- Excess kurtosis < 0 (platykurtic): thinner tails, fewer extremes
Why It Matters for Retirement Planning
Daily and monthly stock returns typically show positive excess kurtosis. This means:
- Market crashes like 2008 (-37%) and 2020 (-34%) are far more likely than a normal distribution predicts
- Extreme rallies also occur more frequently
- Traditional Monte Carlo simulators that assume normal returns underestimate tail risk, potentially giving retirees false confidence
For a whole retirement, the effect on the success rate is usually small when the volatility stays the same (in Retirement Lab's engine, a 4% plan scores 83.3% with a bell curve, 83.1% with fat tails and 82.5% with a downside skew like U.S. history) - the bigger drivers are volatility, the order of returns and how much you spend.
Kurtosis in Retirement Lab
Retirement Lab captures excess kurtosis through the Student's t-distribution, where the degrees-of-freedom (DOF) parameter directly controls tail fatness:
- Higher DOF (e.g., 30): approaches a normal distribution, low excess kurtosis
- Lower DOF (e.g., 5): produces an excess kurtosis of 6 - heavier tails than yearly U.S. stock returns have shown, so it works as a cautious stress test
- DOF = 4: excess kurtosis becomes infinite - extreme events dominate
Retirement Lab offers two settings, moderate (5 degrees of freedom) and extreme (3), both deliberately cautious. Combined with skewness (via the Fernandez-Steel transformation), this captures both the frequency and asymmetry of extreme market events.
How Retirement Lab Addresses This
Retirement Lab captures excess kurtosis through the Student's t-distribution, with a moderate or an extreme setting. Enable fat tails in your Pro simulation to see how realistic kurtosis levels change your success rate compared to thin-tailed normal distribution assumptions.
Frequently Asked Questions
- What is a normal kurtosis value for stock market returns?
- Daily and monthly stock returns show clearly positive excess kurtosis - more extreme moves than a bell curve predicts - while yearly U.S. stock returns since 1928 are close to a bell curve in tail thickness (excess kurtosis near 0) but skewed to the downside. The exact value depends on the asset, the period and how often returns are measured.
- Why does kurtosis matter for retirement planning?
- Higher kurtosis means more frequent extreme market events. With the same average and volatility, fat tails change a 30-year success rate less than most people expect (in Retirement Lab's engine, a 4% plan scores 83.3% with a bell curve, 83.1% with fat tails and 82.5% with a downside skew like U.S. history). They matter more for how bad the worst single years can be.