Patterns and dynamics of retail firm location

Shopping has long been central to urban life, with cities serving as places where demand meets supply for both everyday and specialised goods. However, recent trends in shopping habits (e.g., online shopping), working patterns (e.g., teleworking), and urban development policies (e.g., the 15-minute city) have begun to challenge traditional push-and-pull forces and alter retail firm locations, agglomeration patterns, and consumer behaviour. This dissertation aims to understand how and where shops cluster, and how these patterns evolve as cities grow and develop, providing insights into the complexifying interplay between consumption and urban form. The dissertation is divided into two parts: the first examines observed retail patterns, and the second analyses the dynamics behind these patterns. In the first part, I i) extract retail location data from OpenStreetMap (OSM) for 782 European cities, showing that shops scale super-linearly with city size, while categories scale sub-linearly; ii) analyse intra-urban patterns, calculating walking-distance accessibility under the 15-minute city framework, highlighting that middle-sized Southern European cities are particularly accessible; and iii) assess intra-urban clustering of economic activities, applying the Economic Complexity framework. In the second part, I iv) introduce a theoretical urban model to assess the viability of central retail under online shopping and teleworking preferences; and v) implement this model in an agent-based simulation to explore different location patterns as firms and households interact in space. The results contribute to theoretical and methodological discussions in the field, highlighting how 1) larger cities may offer disproportionately more shops, but middle-sized European cities offer higher accessibility to retail; 2) specialised goods strongly depend on centrality to be viable; 3) consumption generates agglomeration, and strong agglomerations remain key drivers of retail viability, meaning normative models like the 15-minute city must account for these dynamics; 4) both teleworking and online-shopping behaviours act towards less viability of central retail; 5) century-long theoretical discussions (e.g., minimum vs. maximum differentiation) may be reconciled in an agent-based simulation. A proper understanding of the future of urban retail and its relationship to changing consumption trends requires a detailed analysis of existing patterns and a simplified abstraction of complex dynamics, which this thesis attempts to bring to light.