Abstract
Fundamental problems exist with the classical characterisation of agglomeration economies, since such definitions do not reflect the various cost issues on which firms may wish to economise. A lack of understanding of the relationship between the notions of market hierarchies and locational behaviour leads to confusion not only in applied economic interpretation, but more fundamentally in the construction of theoretical location models. In particular, neo-classical location theory can be shown to be crucially flawed as a basis for spatial analysis. This paper therefore attempts to provide an alternative definition of the various types of agglomeration economies such that the various strands of economic theory might be used in a more rigorous manner in the discussion of spatial increasing returns.