A simple asymmetric herding model to distinguish between stock and foreign exchange markets

Drawing on previous work of one of the authors, the paper takes an asymmetric variant of Kirman’s ant model and combines it with an elementary asset pricing mechanism. The closed-form solution of the equilibrium probability distribution allows the specification of a tractable likelihood function for daily returns, which is then employed to estimate the model’s behavioural parameters for a large pool of Japanese stocks. By way of Monte Carlo simulations it is found that most of these markets belong to the same class, which is characterized by a dominance of the stylized noise traders. In contrast, the model assigns a number of major foreign exchange markets to a different class, where on average the majority of agents follows the fundamentalist trading rule. Implications for the tail index are also worked out