Definition
The expected value paradox is the broader class of decision puzzles — of which the St. Petersburg paradox is the established case — in which the conventional expected-value calculation produces a recommendation that no reasonable individual would accept, because the cross-sectional average across possible outcomes diverges from what any single individual will experience along their own path.
Why it matters
The expected value paradox is the umbrella name for a recurring structural problem in decision theory: the natural expected-value criterion, derived from the average over possible outcomes, can produce conclusions that no individual subject to non-ergodic dynamics would or should accept. Ergodicity economics identifies the common source — the divergence of time and ensemble averages — and provides a unified analytical resolution.
How it works
Expected-value paradoxes share a common structure: a gamble or sequence of gambles whose cross-sectional expected outcome — the average across many parallel agents at a moment — is favorable, but whose long-run outcome — the experience of a single agent participating repeatedly — is unfavorable or even ruinous. The St. Petersburg game is one instance; multiplicative betting games with positive expected return but negative long-run growth are another; certain investment strategies under high leverage exhibit the same pattern. The structural source is the same in each case: multiplicative dynamics combined with absorbing barriers (ruin) or fat-tailed outcome distributions, in which the cross-sectional average is driven by extreme outcomes that any single agent realizes vanishingly rarely. Recognizing that the underlying structure is shared is what makes a unified framework possible.
In practice
For an individual evaluating gambles, investments, or strategies with skewed outcome distributions, the expected value paradox is the structural reason to be skeptical of recommendations based purely on expected value. The practical move is to compute or estimate the long-run outcome alongside the expected value, and to recognize when the two diverge enough to change the decision. Real-world examples include over-leveraged investment strategies with positive expected return but high ruin probability, all-or-nothing bets with positive expected value but ruinous loss states, and certain optimization problems whose ensemble-optimal solution is suboptimal for a single agent over time. In retirement income contexts, the analog is that expected-value comparisons across pooled and solo arrangements can hide structurally important differences in path-dependent outcomes.
In the Longevity Standard Framework
Expected value paradox is the category in ergodicity economics of decision-theoretic problems in which ensemble-average reasoning systematically diverges from time-average reasoning for individuals subject to non-ergodic dynamics, and is the broader intellectual context for the Longevity Standard framework's reliance on individual-path treatment rather than expected-utility aggregation. The cost-of-income framework's individual-path orientation is the operational form of treating the participant's long-run outcome as the relevant object of focus rather than an ensemble average over possible outcomes.
Related terms
- St. Petersburg paradox
- Expected utility theory
- Ergodicity
- Non-ergodic system
- Time average
- Ensemble average
- Log utility
- Wealth trajectory