What you described does not necessarily require an agent-based-model (ABM). What you describe could instead be realized in a microsimulation model. Both simulation methods feature a micro and a macro level and interactions between these.
The main differences are that in ABMs, individuals can interact with each other (e.g. in a geographically defined ‘field’), and the macro level will be the sum of micro behaviour and/or effects resulting from said interaction. In micro-simulations, individuals only interact with each other through the macro level (e.g. by supplying something to the “market”) and the macro level results from the aggregation of micro level units (think of aggregate demand/supply).
You can implement microsimulations in all ABM simulation toolkits, but you will not need to implement fields or similar, where agents wander around and interact with their neighbours.
Refer to this table, which classifies simulation models:
Source: Gilbert, N./Troitzsch, K. (2005). Simulation for the Social Scientist. Buckingham: Open University Press, p. 13
This paper by Robert Axelrod contains some general elaborations about the role of simulation is social sciences and advice for researchers.
As @prototoast has mentioned, RePast is the most sophisticated and powerful simulation environment out there. Mastering it requires Java programming skills, and proper display of simulation output can be difficult.
NetLogo, mentioned by @Weijie, is more user-friendly. Its main assets are (1) that the code is interpreted instead of compiled and (2) the easy adjustments of the GUI. This means that models can be built and adjusted very easily. However, the built-in programming language lacks the clarity of Java and is not object-oriented, so larger simulations will soon become messy.
Another alternative might be Ascape/Escape. I haven’t used it so I cannot comment it. It appears to feature model-building wizards of some sort that might replace writing java code yourself.