Agent-Based Modelling

In one sentence

Agent-based modelling (ABM) is a bottom-up simulation approach where many heterogeneous agents follow rules and interact, and macro outcomes emerge from those micro interactions rather than being imposed by an equilibrium solution.

What an ABM contains

An ABM typically specifies:

agents (households, firms, banks, traders) with state variables (wealth, beliefs, inventories),
decision rules (optimizing, heuristics, bounded rationality, learning),
interaction structure (markets, networks, matching, spatial proximity),
institutions (tax rules, monetary policy, contracts),
update dynamics (time steps, shocks, expectations).

    flowchart TD
	  A["Agents<br/>(heterogeneous)"] --> B["Rules / learning"]
	  B --> C["Interactions<br/>(markets, networks)"]
	  C --> D["Emergent outcomes<br/>(prices, volatility, inequality)"]
	  D --> E["Policy experiments<br/>(counterfactuals)"]

Why economists use ABM

ABMs are often used when:

heterogeneity matters (wealth distribution, leverage, firm size),
network effects and contagion matter (financial crises),
equilibrium is hard to compute or may not exist,
learning, norms, and bounded rationality are central.

Strengths and limitations

Strengths

can represent realistic institutional detail and frictions,
produces rich dynamics (nonlinearities, fat tails) without assuming them,
supports “what-if” policy experiments when analytical solutions are intractable.

Limitations

calibration/validation can be difficult (many degrees of freedom),
identification is hard (different rule sets can fit the same facts),
results can be sensitive to modelling choices (requires transparency and robustness checks).

ABM vs DSGE (quick contrast)

DSGE models typically impose optimizing behavior and equilibrium conditions; ABMs often allow out-of-equilibrium dynamics and rule-based behavior. In practice, they can be complementary: ABM can explore mechanisms and nonlinearities; DSGE can offer clearer identification and welfare interpretation under strong assumptions.

Game Theory: The study of mathematical models of strategic interaction among rational decision-makers.
Computational Economics: The application of computational methods to economic problems.
System Dynamics Modelling: A simulation approach focused on aggregate stocks/flows and feedback loops (typically more top-down than ABM).
Complex Systems: Systems with many interacting parts where aggregate behavior is not obvious from individual components.

Quiz

### What is the primary purpose of Agent-Based Modelling? - [x] To simulate the decisions and interactions of individual agents - [ ] To aggregate economic data - [ ] To develop linear regression models - [ ] To calculate national GDP > **Explanation:** ABM is primarily used to simulate the decision-making processes and interactions of individual agents within an economic system. ### Which of the following fields can benefit from ABM? - [x] Urban planning - [x] Epidemiology - [x] Financial market simulations - [x] Social behavior modeling > **Explanation:** ABM has versatile applications across multiple disciplines, including urban planning, disease spread modeling, financial markets, and social behaviors. ### True or False: ABM uses a top-down modeling approach. - [ ] True - [x] False > **Explanation:** ABM employs a bottom-up approach, focusing on individual agents and their interactions rather than a holistic top-down approach. ### Who was a key figure in demonstrating segregation dynamics using early forms of Agent-Based Models? - [ ] John Nash - [ ] Adam Smith - [ ] Joseph Schumpeter - [x] Thomas Schelling > **Explanation:** Thomas Schelling was known for using an early form of agent-based model to simulate segregation dynamics. ### What kind of interactions can be efficiently simulated by ABMs? - [x] Complex Interactions - [ ] Linear trends - [ ] Simple aggregations - [ ] Monotonic progressions > **Explanation:** ABMs are particularly effective at simulating complex interactions among diverse agents. ### Which institution played a pivotal role in promoting ABM in the 1980s? - [x] Santa Fe Institute - [ ] World Bank - [ ] United Nations - [ ] Harvard University > **Explanation:** The Santa Fe Institute was instrumental in the development and promotion of ABM during the 1980s. ### Can ABM handle heterogeneous agents and decision rules? - [x] Yes - [ ] No > **Explanation:** Yes, ABM excels at incorporating heterogeneous agents with diverse decision-making criteria and behaviors. ### What does the “utility” term in ABM context refer to? - [x] The satisfaction or happiness derived by an agent - [ ] The total revenue of a firm - [ ] The cost of production backfill - [ ] The national income > **Explanation:** In the context of ABM, "utility" refers to the satisfaction or happiness that a consumer or agent derives from their consumption choices. ### In ABM, do the simulation results affect future decision periods? - [x] Yes - [ ] No > **Explanation:** Yes, the simulation results impact the environment and conditions for subsequent decision periods, contributing to the dynamic evolution of the system. ### True or False: Game Theory often uses computational simulations similar to ABM. - [ ] True - [x] False > **Explanation:** While Game Theory and ABM both analyze individual decisions and interactions, Game Theory typically employs analytical methods rather than computational simulations used in ABM.