We propose a subsidy-based control law of a multi-agent system that consists of many selfish and heterogeneous agents under incomplete information on the criterion of each agent. The system's behavior is modeled by multipopulation replicator dynamics where the payoff functions of the populations are different from one another and correspond to the criteria of the agents who belong to the populations. In general, several equilibrium points exist in the replicator dynamics. A controller called a "government" attempts to stabilize the desirable equilibrium points of multiagent systems by offering a subsidy to the agents. A previous study assumed that the government determined the subsidy on the basis of the population that each agent belonged to. In general, however, the government cannot identify the members of each population. In this technical note, we assume that the government observes the action of each agent and determines the subsidy on the basis of the observed action profile. Then, we model the controlled behavior of the system using replicator dynamics with feedback. We derive the stabilization condition of the target equilibrium point for two special cases.