The Equity Premium Puzzle: An Application of an Agent-Based Evolutionary Model

We describe an agent-based model of a financial market with a stock and a bond. Agents compete in repeated rounds, decide whether to acquire costly information and can pick one of 16 strategies to allocate their investments, under evolutionary pressure driven by the comparison of the realized short-term revenues from trading. We show that, while in- formed traders survive in some cases, the equilibrium shares are strongly biased in favor of strategies that make little use of information and sys- tematically overestimate the riskiness of the stock. As a consequence, the majority of the population ends up in buying fewer stocks than would be otherwise expected or deemed rational. This evolutionary dynamics offers a novel way to explain the equity pre- mium puzzle first described by Mehra and Prescott (The equity pre- mium: A puzzle. Journal of Monetary Economics 1985), according to which it’s hard to find reasons for the widespread lack of investment in risky assets. Evolution based on a straightforward comparison of rev- enues is a simple and cognitively appealing avenue to reach a population of traders using (over-)cautious strategies to curb the risk of long-term “financial extinction”. Simulations run in NetLogo also demonstrate that very little information may be used in noisy markets or when the cost of information is substantial.