Ergodicity

Ergodicity is the property of a dynamical system whereby time averages equal ensemble averages: the long-term behavior of a single trajectory reveals the statistical properties of the entire state space. A system is ergodic if, given enough time, it visits all accessible states with a frequency proportional to their probability in the equilibrium distribution. This is not a universal property; it is a specific structural feature that must be proven or assumed.\n\nThe concept originated in statistical mechanics, where the ergodic hypothesis — the assumption that a thermodynamic system explores all of its phase space — underlies the connection between microscopic dynamics and macroscopic thermodynamics. But ergodicity has migrated far beyond physics. In Markov chains, ergodicity is the condition that guarantees convergence to a unique stationary distribution. In economics, it separates systems where individual trajectories reflect population averages from systems where individual outcomes diverge irreversibly from aggregate behavior.\n\nThe philosophical significance of ergodicity is easily overstated. A system can be ergodic yet mix so slowly that no real observer will ever see the equilibrium distribution. The mixing time — the rate at which a system approaches statistical uniformity — is often more practically relevant than ergodicity itself. Ergodicity is a limit property; mixing is a rate property, and in most real systems, rates matter more than limits.\n\nThe ergodicity debate in economics — whether human systems are ergodic or not — is not a debate about mathematics. It is a debate about whether individuals can be treated as interchangeable draws from a population distribution, and the answer, in any system with memory, compounding, and path dependence, is no.\n\n