Risk Aversion in Non-Ergodic Systems

TL;DR

This paper demonstrates that in non-ergodic systems, agents' risk aversion emerges from evolutionary dynamics, and their subjective probabilities align with risk-neutral probabilities, challenging traditional views on risk preferences.

Contribution

It introduces a novel approach linking evolutionary simulation to the derivation of risk-neutral probabilities, removing the need for assumed risk aversion.

Findings

Subjective probabilities evolve to favor bad events over good ones.

Risk-neutral probabilities can be derived from evolutionary dynamics.

Risk aversion disappears when agents focus on time averages instead of ensemble averages.

Abstract

We show in a simulation when economic agents are subject to evolution (random change and selection based on the success in the estimation of the result of the gamble) they acquire risk aversive behavior. This behavior appears in the form of adjustment of their estimation of probabilities when calculating the expected value (ensemble average). It means that their subjective probabilities evolve in such a way that economic agents tend to assign lower probabilities to "good" events and higher probabilities to "bad" events. These subjective probabilities can be derived analytically by assuming that economic agents care about time average, not the ensemble average. Probabilities calculated based on this assumption are equal to the probabilities we get in evolutionary simulation. Furthermore, it appears that these subjective probabilities are equal to risk-neutral probabilities in mathematical finance. Hence, by taking into account that the environment in which economic agents operate is non-ergodic, we are able to calculate risk-neutral probabilities that are consistent with modern finance theory but are derived in a conceptually different way. It means that when we assume that agents try to predict time average, not ensemble average, the need for the concept of risk aversion disappears, since there is no distortion of subjective probabilities. Evolutionary simulations are quite a general method that can be applied for the determination of relevant measures of the outcome of a gamble under various conditions.