Social Interactions. A Formal Approach to Feedback-Based Evolutionary Local Learning

This paper discusses the usefulness of SI models in explaining feedback-based local learning. The literature on evolutionary economic geography gives feedbacks a major role in shaping local systems, nevertheless the econometric approaches to local systems are mainly focused on estimating spillovers. Although some works have already proposed to use social interactions models to approach local systems, they have not focused on feedbacks processes involved in local interactions neither have estimated the social multipliers implied. This paper proposes that feedbacks dynamics can be capture estimating social multipliers implicit in social interactions models and by distinguishing between different types of interactions that can give rise to local spillover. In this regard tree different types of interactions are distinguished: (i) pure spillovers or common shocks, (ii) self-selection –choosing the location according the characteristics of co-located firms-, and (iii) pure social interactions – when the behavior of co-located firms affect individual behavior-. Feedbacks will be present in the last two types. This distinction will be used for detecting feedbacks from pure spillovers, in which most of the literature has focused. Although the objective of this paper is to explore a methodology and to discuss its scope, in the last section the paper offers an estimation of social multipliers in local systems made up by SMEs that share the same industry and geographical localization. This estimation is run in a microdatabase compound by 1233 Argentinean SMEs with information on productivity and innovation behavior among other variables for the period 2006 -2008. The results show positive and significant social multipliers and that cannot be the result of pure spillover. Therefore, the results offer statistical evidence that support the hypothesis of feedbacks in local interactions.