Strategic decision-making in multi-agent markets: The emergence of endogenous crises and volatility

Abstract

Traditional economic frameworks are built upon perfectly rational agents and equilibrium outcomes. However, during times of crises, these frameworks prove insufficient. In this thesis, we take an alternative perspective based on "Complexity Economics", relaxing the assumption of perfectly rational agents and allowing for out-of-equilibrium dynamics. While many contemporary approaches explain crises and non-equilibrium market phenomena as the rational reaction to external news, the emergence of endogenous crises remains an open question.

We begin addressing this question by demonstrating how a multi-agent model of heterogeneous boundedly rational agents acting according to heuristics can reproduce and forecast key non-linear price movements in the Australian housing market, during boom and bust cycles. In order to provide foundations for such heuristic-based reasoning, we then propose a novel information-theoretic approach, Quantal Hierarchy, for modelling limitations in strategic reasoning, demonstrating how this convincingly and generically captures the decision-making of interacting agents in competitive markets outperforming existing approaches. In addition, we demonstrate how a concise generalised market model can generate important stylised facts, such as fat-tails and volatility clustering, and allow for the emergence of crises, purely endogenously. This thesis provides support to the interacting agent hypothesis, addressing a crucial question of whether crisis emergence and various stylised facts can be seen as endogenous phenomena, and provides a generic method for representing strategic agent reasoning.