More

Networks & Agent Based Models

People are different, in the same way that no two businesses, industries or countries are exactly like. Diversity in behaviours, knowledge, opinions and actions is an obvious feature of everyday life, however this heterogeneity is not reflected in most economic models and assumptions.

 

As well, we know that in many complex social and economic contexts, decision makers often pay attention to each other, looking to the behaviours and actions of others to inform choices between alternative courses of action. This is particularly so when information about the problem itself is imperfect or when individuals are unable to optimally process even the information that is available – in these complex scenarios, taking advice or using recommendations from others makes sense.

 

Agent-based modelling (ABM) provides a way to move forward from the assumptions of uniformity and perfect information by allowing analysis of the interaction between heterogeneous individuals. ABMs specify simple rules of behaviour for agents – which can represent people, firms or nations – and then simulate the system level outcomes that emerge as agents interact.

 

This ‘bottom-up’ approach to modelling avoids the need to make assumptions about equilibrium states, instead letting the system evolve through simulated time. This non-equilibrium approach is especially useful for modelling phenomena characterised by bubbles, herd behaviour and tipping points - in other words, whenever it is more useful to understand how the system is likely to change dynamically, rather than its properties under a theoretical, long-run steady state.