Abstract
This paper deals with the constraints and risks of learning in different types of spatial concentration of related industries and firms. We aim at a better understanding of what makes the difference between local lock-in on the one hand and ongoing creation of novelty on the other. To achieve this purpose, we use Nonaka and Takeuchi's (1995) treatment of knowledge conversion processes and Nooteboom's (2000) cycle of discovery. Hence, we are able to clarify the concept and nature of learning, which in turn provides a basis for specifying different learning effects of two prototypes of spatial concentration: Marshallian and dynamic industrial districts. We show that these two types of industrial districts have multiple, different, and complementary functions in terms of knowledge conversion and knowledge creation. Hence, we can explain why spatial concentration can have positive and negative effects for learning and innovation, and how lock-in can be avoided.