A Causal GeoAI Framework for Capturing the Unearned Increment to Self-Finance Public Infrastructure

Abstract

The foundation of a sustainable urban economy relies on a circular value mechanism where the spatial wealth generated by the city is recaptured to finance its own infrastructure. However, current public financing and static land management practices fail to equitably measure and capture the unearned value increment created by infrastructure investments. Because the complex and non-linear nature of urban market dynamics makes it difficult to isolate the causal impact and spatial heterogeneity generated by these investments, this massive uncaptured rent fuels private speculation and undermines the financing capacity of local governments. As an analytical solution to this chronic value capture gap and measurement problem, this study introduces an innovative spatial decision intelligence framework oriented towards Land Value Capture (LVC). Tested in the high-density metropolis of Istanbul, the methodology establishes an integrated structure coupling causal machine learning algorithms—which isolate the pure effects of public amenities on real estate values—with an AHP-TOPSIS decision model that optimizes these causal rent simulations along the axes of social resilience and spatial justice. Empirically, LightGBM outperformed traditional hedonic models (R²=0.67). Causal forests revealed facility-specific spatial externalities: green spaces generated localized premiums (α=0.411), whereas large healthcare and educational facilities induced widespread value suppression. Consequently, while unidimensional decisions focused solely on revenue maximization conflict with societal benefit, the proposed multidimensional mechanism successfully strikes a balance between economic rent and spatial justice. Overall, this hybrid framework transforms public investment site selection from a passive expenditure into a strategic tool for building a rational, equitable, and self-financing urban economy.