Switching rates and the asymptotic behavior of herding models

Markov chains have experienced a surge of economic interest in the form of behavioral agent-based models that aim at explaining the statistical regularities of financial returns. We review some of the relevant mathematical facts and show how they apply to agent-based herding models, with the particular goal of establishing their asymptotic behavior because several studies have pointed out that the ability of such models to reproduce the stylized facts hinges crucially on the size of the agent population (typically denoted by n), a phenomenon that is also known as n-dependence. Our main finding is that n-(in)dependence traces back to both the topology and the velocity of information transmission among heterogeneous financial agents.