Productivity determinants are a central research topic. Among these determinants, the impact of territorial resources (especially at sub-national scales) has been under-researched and relatively neglected by mainstream scientific research. Moreover, many productivity studies deal with it at an aggregate level, which may lead to ecological errors in the conclusions related to public policies and regional development. This doctoral research aims to answer three questions to tackle these gaps and issues. To start with, what are the territorial resources that significantly influence firms’ productivity? Methodologically, can we avoid the bias of the ecological error influencing interpretations and conclusions outlined by studies conducted at aggregated levels? And about policy mix, what specific lessons can we learn from improved understanding of the relationship between territorial resources and firm productivity? The thesis was developed in three research areas to address the research questions. Firstly, two systematic literature reviews were conducted to identify the leading productivity indicators, the most appropriate statistical methods and the relevant territorial resources. Then, the spatial structures of several productivity indicators were identified using spatial autocorrelation indexes. Spatial analysis and cartography are suitable tools to identify geographies of firm productivity, but they are currently underused. Finally, based on the spatial structures and the literature, we developed an explanatory model of Walloon firms’ Total Factor Productivity (TFP) by territorial resources. The results underline agglomeration economies’ central role: an imbalance does exist between economies and diseconomies in the major Walloon urban areas. The economies (linked to urban size) are too small by comparison with the structural difficulties related to the industrial decline of these urban areas. In addition, labour markets of nearby metropolises like Brussels or Luxembourg seem to put pressure on Walloon local ...