The socioeconomic phenomena as tax evasion in agent-based community can be studied numerically through the Zaklan model (ZM). The ZM has been con_gured as an excellent tool to control the tax evasion uctuations in communities with agents or individuals that can be considered tax evaders or honest. The dynamic of temporal evolution of the ZM was studied initially via the equilibrium Ising model (IM), non-equilibrium Majority-vote model (MVM) and recently via a three-state kinetic agent-based model. Here, through Monte Carlo simulations, we study the problem of the tax evasion uctuations using a discrete version of the kinetic model of opinion dynamics (BCS) in the dynamic of the temporal evolution of the ZM. Unlike previous models cited here in the BCS model, the interactions between agents occur in pairs. Then, we found that the BCS model is as e_cient as the IM and MVM in controlling the tax evasion uctuations. This control is even better when we use strong punishment values k even for low audit probabilities pa.