The Geography of the Green Transition under NextGeneration EU. A Text-Learning approach

This paper investigates the geography of the green transition by moving beyond aggregate expenditure to analyse the project-level composition of green investments under the Italian National Recovery and Resilience Plan. We classify 92,000 funded projects into nine green trajectories using a policy-anchored Transition Trajectory Classification. To examine how the distribution of these trajectories is associated with contextual characteristics, we use multinomial logit models incorporating spatial dependence structures. Results show that investments reflect territorial capacity and pre-existing conditions: innovative-intense projects cluster in metropolitan areas, peripheral and less developed municipalities focus on restoration and adaptation, and some of these areas remain completely excluded. Green territorial divide thus operates through unequal access to funding and qualitative segmentation of transition trajectories.