Smoothing the Edges: Reconceptualizing Space and Dealing With Modifiable Areal Unit Problems in (Un)Related Variety Research

Abstract

Related variety studies in Economic Geography reveal regional diversification mechanisms for regional development, but often overlook geographic fundamentals. By relying on administrative units, the studies may fail to account for spatial continuity and interdependence, which can lead to Modifiable Areal Unit Problems. In this regard, this article introduces an alternative method ‘smoothing the edges’ as proof of concept to strengthen spatial conceptualization. Instead of using administrative units, we construct high-resolution grid cells and define Local Economic Environments (LEEs) around them to calculate economic factors. LEEs capture the conditioning economic context to which each grid cell is exposed. We compare Ordinary Least Squares regression outcomes across three LEE scales, equivalent to NUTS 2, NUTS 3, and municipality levels, and examine how (un)related variety effects behave across scales under the new conceptual framework. We apply two stylized facts from the literature: (Un)Related variety associates with (1) industrial specialization, and with (2) employment growth. A case study with Dutch establishment microdata LISA reveals that effects of (un)related variety are sensitive to scale, particularly in employment growth analysis. These findings highlight the importance of understanding contextual settings, which is critical in informed policy making.