This paper examines how local public amenities are distributed across neighbourhoods within cities. Using comprehensive data on libraries, parks, pools, leisure centres, and infrastructure investments in England (2005–2024), I document a striking U-shaped relationship between neighbourhood affluence and public amenity access: the poorest and wealthiest neighbourhoods have the best access, while middle-income areas are systematically underserved. To address endogeneity from household sorting and amenity capitalisation, I exploit quasi-experimental variation from the 2014 Stamp Duty Land Tax reform, which generated discontinuous changes in transaction costs across the house price distribution. Instrumental variable estimates confirm the U-shaped pattern. Investigating mechanisms, I show that poor neighbourhoods exhibit high demand for public amenities, while rich neighbourhoods have greater political power. In contrast, private amenities like restaurants or bars concentrate exclusively in affluent areas. A quantitative spatial model with endogenous public amenity provision rationalises these patterns through political economy: local governments balance redistribution toward needy neighbourhoods and capture by politically powerful wealthy neighbourhoods, leaving the middle class behind. The findings reveal a novel form of spatial inequality within cities and have implications for local public finance, urban policy, and quantitative spatial models.

Climate action requires significant public and private sector investments to achieve meaningful reductions in carbon emissions. This paper documents that austerity, coupled with a lack of (digital) skills in (local) government, may have been a significant barrier to delivering climate action in the form of retrofitting. Decomposing heterogeneity in estimated treatment effects of a large scale energy efficiency program rolled out through a regression discontinuity design in the early 2010s, we find that both the extent of local budget cuts and poor digital connectivity may be responsible for up to 30% fewer retrofit installations that counterfactually would have taken place.