Artificial Economics can be defined as ″a research field that aims at improving our understanding of socioeconomic processes with the help of computer simulation″.[1] Like in Theoretical Economics, the approach followed in Artificial Economics to gain understanding of socioeconomic processes involves building and analysing formal models. However, in contrast with Theoretical Economics, models in Artificial Economics are implemented in a programming language so that computers can be employed to analyse them.
Concretely, the method followed in Artificial Economics to analyse formal models most often comprises two stages: 1) deductive generation of samples, and 2) inductive inference of general patterns.[2] [3]
Thus, using this computer simulation approach, the data is produced by the computer using strict deduction, but the general patterns about how the rules of the model transform the inputs into the outputs are inferred using generalisation by induction.
The benefit of using the computer simulation approach described above (vs. pure logical deduction only) is that it enables the exploration of (formal) models that are –currently– intractable using the most advanced mathematical techniques. This is so because the set of assumptions that can be investigated using computer simulation is not limited by the strong restrictions that mathematical tractability imposes. This point is particularly important in the study of socioeconomic processes, which –due to its complex nature– are oftentimes difficult or impossible to address adequately using a purely deductive approach only. The strictly deductive approach often requires so many simplifications to ensure mathematical tractability that the correspondence between the real world and the model assumptions turns out disappointingly weak.
Some of these simplifications have been outlined in the left column of the table below, together with some of the features that can be explored using the Artificial Economics approach (right column).[6]
| Traditional restrictions imposed to ensure mathematical tractability | Features that can be explored with Computer Simulation (Artificial Economics approach) | |
|---|---|---|
| Representative agent or a continuum of agents | Explicit and individual representation of agents (agent-based modelling) | |
| Rationality (and sometimes common knowledge of rationality) | Adaptation at the individual level (learning) or at the population level (evolution). Satisficing | |
| Local and asymmetric information | ||
| Focus on out-of-equilibrium dynamics | ||
| Stochasticity | ||
| Top-down analysis | Bottom-up synthesis | |
| Random or complete networks of interaction | Arbitrary (and potentially endogenous) networks of interaction | |
| Minor role of physical space | Explicit representation of physical space | |
| Infinite populations | Finite populations | |
| Preference for uniqueness of solutions | Path dependency and historical contingency |
The differences in the type of assumptions investigated using the strictly deductive approach only and those investigated in Artificial Economics are so fundamental that some scholars[7] see these differences as the defining features of Artificial Economics. Other scholars find that the distinctive characteristic of Artificial Economics is methodological, i.e. the use of the computer simulation approach. The fact that models in Artificial Economics are implemented in a programming language (rather than expressed as a set of equations) is not considered substantial since any model implemented in computer code can be expressed as a well-defined mathematical function.[8] [9] [10]
One of the aim of these conferences is to favour the meeting of people and ideas coming from two communities of scientists –computer science and economics– in order to construct a more structured multi-disciplinary approach.[11] Proceedings of every conference in the series have been published as a volume in the Lecture Notes in Economics and Mathematical Systems Springer series.