Abstract
To create a valid model of a social or economic system one must consider a lot of parameters, conditions and restrictions. Systems and, consequently, the corresponding models are proved to be very complicated. The problem of such system model engineering can’t be solved only with mathematical methods usage. The decision could be found in computer simulation. Simulation does not reject mathematical methods, mathematical expressions could become the foundation for a computer model. In these materials the set of agent-based computer models is under discussion. All the set models simulate productive agents communications, but every model is geared towards the specific goal, and, thus, has its own algorithm and its own peculiarities. It is shown that computer simulation can discover new features of the agents’ behavior that can not be obtained by analytical solvation of mathematical equations and thus plays the role of some kind of economic microscope.