The Austrian Economics Model is an agent-based model (ABM). Agent-based modeling is a type of computer simulation based on “agents” with independent behavior that interact with other agents and their environment. Though in general, agents in agent-based modeling can be anything with individual behavior, such as molecules, plants, or animals, agents in the Austrian Economics Model represent human beings that pursue their own economic goals and make their own economic decisions.
In particular, the Austrian Economics Model is inspired by the Sugarscape model of Joshua Epstein and Robert Axtell. The Sugarscape model was the first large-scale agent-based model. Starting with agents that move around a grid world to collect “sugar” in order to survive, Epstein and Axtell were able to add more complex phenomena such as trade and wealth distribution, war, evolution, and the transmission of diseases.
The Austrian Economics Model also begins with agents that move around in a grid world to collect natural resources. From these natural resources, agents produce more complex goods, trade with each other, and choose different strategies to obtain their economic goals. The idea is to create believable artificial economies which capture crucial aspects of real life economics, especially through the insights of the Austrian school.
There are three elements of agent-based modeling that are particularly important here: heterogeneity, imperfection, and a cognitive architecture.
Heterogeneity. The agents in the model are heterogeneous; they have different physical conditions, subjective preferences, and natural skills and capacities. Austrian economics disagrees with forms of economic modeling that assume homogeneous populations. In fact, Austrian economics insists that the differences between individuals are what drive an economy. Individuals exchange items because either party values what they receive more than what they give in exchange; the specialization of labor and services is encouraged by the fact that some people have talents that others do not; entrepreneurs are creative individuals who see opportunities that others fail to see. It is important to simulate an economy where individuals are different.
Imperfection. It will surprise non-economists to know that most mathematical economic models use perfectly rational and omniscient consumers and producers. This is because such constructs behave predictably and fit smoothly into the equations of economists. However, such consumers and producers do not exist in real life. Austrian economics, on the other hand, recognizes the imperfection of individuals. They have irrational desires, imperfect knowledge, and make mistakes about the future. The Austrian Economics Model uses such imperfect agents. These are agents that can sometimes fail, but can also learn from their failures.
Cognitive Architecture. The core of the model is the implementation of a BDI (Belief-Desire-Intention) cognitive architecture for agents. The minds of agents are structured according to what they know/believe, what they want/desire, and the plans that they employ/intend to obtain their desires. The BDI cognitive architecture is intuitive because it captures how human beings think they think. This has made it one of the most successful cognitive architectures in simulation with implementations in different programming languages and platforms. In the Austrian Economics Model, the BDI framework dictates a wide range of actions that an agent can take, from negotiating prices with a seller to creating a brand new product.
In addition to the three elements from agent-based modeling just mentioned, the model also relies on procedural generation and a certain degree of abstraction. The world is not predetermined but procedurally generated. The geography, attributes of agents, kinds of goods, and the possibilities involved in combining these goods are procedurally generated once a simulation is run. This leads to different configurations for economies. Also, the goods in the artificially generated economies do not directly correspond with real world goods; there are no “bananas” or “automobiles.” Instead, goods are artificial constructs with certain properties and combinatorial possibilities. For example, good A3 may be something that increases an agent’s health (it is a kind of food), which can be combined with good B2 using skill 7S to produce good C5 which many agents find attractive (perhaps a kind of jewelry). This degree of abstraction allows for different interpretations and also allows the Austrian Economics Model to not be bound to a specific economic or cultural history. Rather, it is able to come up with different kinds of artificial economies.
The Austrian Economics Model is programmed using Godot, a free and open-source game engine under the permissive MIT License. The data analysis is done with Python. For screenshots of the current version of AEM, click here.