Several teaching methodologies have been developed to enhance student engagement in a classroom, and a reward structure to evaluate and encourage students is a standard component. However, rewarding a student not only motivates the recipient but also raises social comparison in a classroom. While this concern has long been expressed, empirical studies are limited due to the lack of variety in realistic cases. In this work, we construct an artificial society to investigate the utilities of different reward structures undergoing students’ social comparisons. The proposed complex system follows Adams's equity theory that student agents who feel unfair will reduce their engagement levels and vice versa. Through a series of simulations, we propose three inferences from the findings. First, rewarding students proportional to their classroom performance is adequate to obtain a high engagement level of most students, whereas some students’ achievements considerably fall behind. Secondly, rewarding students by their engagement levels can be seriously harmful to the performance of high-ability students in our model. Lastly, if students can be guided to compare with peers of similar abilities, equally rewarding students then becomes the optimal scheme for both students and educators. Our theoretical approach reveals the potential of computational modeling on the analysis of complex student behavior and brings in further validation issues on the other hand.