2008/26 LEM Working Paper Series

Pseudo-NK: an Enhanced Model of Complexity

Marco Valente
NK model, Simulation models, Complexity, Interactions

  JEL Classifications
C15, D20, D83, L23, O31, O32

This paper is based on the acknowledgment that NK models are an extremely useful tool in order to represent and study the complexity stemming from interactions among components of a system. For this reason NK models have been applied in many domains, such as Organizational Sciences and Economics, as a simple and powerful tool for the representation of complexity. However, the paper suggests that NK suffers from un-necessary limitations and difficulties due to its peculiar implementation, originally devised for biological phenomena. We suggest that it is possible to devise alternative implementations of NK that, though maintaining the core aspects of the NK model, remove its major limitations to applications in new domains. The paper proposes one such a model, called pseudo-NK (pNK) model, which we describe and test. The proposed model appears to be able to replicate most, if not all, the properties of standard NK models, but also to offer wider possibilities. Namely, pNK uses real-valued (instead of binary) dimensions forming the landscape and allows for gradual levels of interaction among components (instead of presence-absence). These extensions provide the possibility to maintain the approach at the original of NK (and therefore, the compatibility with former results) and extend the application to further domains, where the limitations posed by NK are more striking.

download pdf