In this paper, two different versions of the agent-based DSGE (dynamic stochastic general equilibrium) model are studied in comparison The first version is the mesoscopic modeling of market sentiment using the Brock–Hommes adaptive belief system (ABS). The second version is the microscopic modeling of market sentiment by applying the Ising model to different social networks. The issue is to examine whether the distribution of market sentiment generated by the ABS machine can emerge from some kinds of local mimetic interactions, and hence the macroeconomic behavior of the two versions will be essentially the same. Our simulation results show that it is rather hard to have the equivalence of these two versions in the Kolmogorov–Smirnov sense. Hence, directly incorporating social networks and social interactions into microscopic modeling has its own values and may not be replaced or simplified by the mesoscopic counterpart.
Research in International Business and Finance, 2012