Dots to dots: a general methodology to build local indicators using spatial micro-data

Abstract

Empirical studies in regional science have so far largely relied on discrete conceptualizations of space and aggregated metrics, which do not take into consideration spatial heterogeneity and variability at the micro-level. This paper explores the use of these indicators when dealing with observations at the subregional level, based on micro-data sets that impose the conceptualization of spatial interactions in a continuous and multidirectional space. We propose a general methodology to build local indicators for spatial micro-data sets. Based on distance matrix and matrix calculations, some classical indices of specialization and diversity are extended to their local counterparts to explore the full spatial heterogeneity and variability of space. The methodology is applied to 9,839 establishments covering all economic sectors in the Lower St-Lawrence region (Quebec, Canada). We find that the distribution of the local indicator varies significantly with distance, which suggests that the effects of specialization or diversity are not constant over space. Treating space as continuous may become of prime importance, given that more individual data sets are now available, combined with the fact that the performance of microcomputers is still improving.