Summary of key model parameters.

This paper applies statistical mechanics to investigate wealth distribution in binary interactions between two groups of agents. Using an exchange rule with non-zero expected random variables and non-Maxwellian collision kernels, we consider the case that wealth distribution is affected by the wealth replacement rate, trading rate, market risk and the proportion of steady-state wealth distributions of two groups of agents. The decrease of market risk and the increase of the wealth replacement rate and trading rate are conducive to the equalization of wealth distribution, and high proportion of steady-state wealth distributions of two groups of agents narrows disparities in group 1 but worsens them in group 2 under certain conditions. We verify our conclusions by numerical experiments.