Treatment, Exposure, and Spatial Counterfactuals in Regional Policy Evaluation: A Framework for Impact Analysis With Spatial Spillovers

Regional policies often affect untreated but spatially exposed units, making standard difference‐in‐differences comparisons difficult to interpret. This paper develops a treatment‐exposure framework for regional policy evaluation with spatial exposure and interference. It shows that two‐way fixed effects estimators that omit exposure conflate the direct treatment effect with exposure components among currently untreated units and treated‐exposure components whose signs depend on the exposure design. A decomposed specification separates direct effects, exposure among currently untreated units, and treated‐exposure components, conditional on a maintained exposure mapping specified before estimation. The paper translates these conditions into an identification protocol for selecting, diagnosing, and stress‐testing exposure mappings. The contribution is methodological and operational: the paper synthesizes insights from interference, spatial DiD, and spatial econometrics into a protocol for applied econometric practice, while clarifying how decomposed exposure specifications and event‐study diagnostics shape the interpretation of spatial policy effects. The protocol emphasizes that exposure‐mapping sufficiency cannot be established from outcomes alone: alternative matrices, higher‐order exposure terms, and pre‐trend diagnostics can challenge specific implications of the maintained mapping, but they do not prove that the scalar exposure index is sufficient. An application to London's Ultra Low Emission Zone estimates a direct reduction of 6.6% and an implied currently untreated exposure reduction of 6.2% at the mean positive exposure level under the benchmark 0–3 km exposure mapping. A dynamic decomposition shows that this currently untreated exposure component has different meanings across expansions: anticipatory adjustment among future‐treated stations in the 2019 stack and spatial exposure or network‐mediated reallocation after the 2021 expansion. Spatially bounded evaluations restricted to directly treated units may miss exposure channels and therefore provide an incomplete basis for assessing the affected population and the policy's costs and benefits.