I suppose there could be multiple ways to model the gender pension gap. So really, it would come down to the modeling approach and assumptions that you are using. For example, the spread of a virus could be modeled with differential equations (e.g., SIR model), or agent based models. Often times, differential equations are used when mathematical abstractions are a good approximation and you can describe how relevant quantities are changing. ABMs might be better when there is significant heterogeneity, or discreteness which is difficult to boil down to a set of equations.
Without knowing much about your specific model, its difficult to say whether Agents.jl would be a good fit. Nonetheless, its very flexible and has lots features, examples, and documentation.