Definition
A fat-tailed distribution is a probability distribution in which extreme outcomes — very large or very small results — happen more often than common intuition or the bell-curve normal distribution would suggest, so that rare large events are not as rare as they look.
Why it matters
Many real-world outcomes — investment returns, longevity in old age, claim sizes in insurance, sizes of catastrophic events — exhibit fat tails, and the fatness of the tail has structural implications that thin-tailed analysis misses. Under fat tails, sample averages converge slowly to true averages, expected values can behave erratically, and ensemble averages over short observation windows can be systematically misleading. For ergodicity economics, fat tails combined with multiplicative dynamics produce particularly severe divergences between time and ensemble averages.
How it works
The defining feature of a fat-tailed distribution is how slowly the probability of extreme outcomes drops off as the outcomes get more extreme. In a normal (bell-curve) distribution, the probability of an outcome multiple standard deviations away from the mean drops off very quickly — extreme outcomes are effectively never seen. In a fat-tailed distribution, the probability of extreme outcomes drops off more slowly, so that extremes the normal distribution would treat as essentially impossible occur with meaningful frequency. The fatter the tail, the more often extremes occur — and the more a small number of extreme realizations can dominate any sample's average. Under multiplicative dynamics, fat-tailed period factors compound into wealth trajectories where the long-run experience of any single path can be dominated by a small number of extreme events that the cross-sectional average smooths away.
In practice
For an individual reasoning about long-horizon outcomes — investment returns, longevity, sequence-of-returns risk in retirement — recognizing that the underlying distribution has fat tails is the structural recognition that extreme outcomes are not freak occurrences but a regular feature of the dynamics. The practical move is to ask whether historical sample statistics like mean returns and standard deviations are reliable summaries of the underlying distribution, or whether the distribution has tail behavior the sample has not yet revealed. For longevity specifically, the survival distribution has a fatter right tail than common intuition acknowledges — the probability of an individual living past 95, 100, or 105 is higher than most planning rules account for, which is one structural reason why solo drawdown's choice of planning age is particularly consequential.
In the Longevity Standard Framework
Fat-tailed distribution is the established formulation in ergodicity economics of probability distributions whose extreme tails make ensemble-average reasoning particularly unreliable for individual-path outcomes, and is the structural feature of the underlying mortality and returns distributions on which the Longevity Standard framework's actuarial work rests. The framework's Gompertz mortality assumption produces a survival distribution with structurally important right-tail behavior — the fraction of the focal individual's survival probability concentrated in ages past life expectancy is what makes pooled and transferred-risk arrangements valuable at all, by spreading the cost of that right tail across the pool. Solo drawdown's planning-horizon risk is the individual-path manifestation of the same fat-tail structure.
Related terms
- Tail risk
- Absorbing barrier
- Ruin probability
- Non-ergodic system
- Multiplicative dynamics
- Longevity tail risk
- Wealth trajectory
- Time average