A causal GeoAI framework for capturing the unearned increment to self-finance public infrastructure
Sustainable Cities and Society, cilt.148, 2026 (SCI-Expanded, Scopus)
- Yayın Türü: Makale / Tam Makale
- Cilt numarası: 148
- Basım Tarihi: 2026
- Doi Numarası: 10.1016/j.scs.2026.107595
- Dergi Adı: Sustainable Cities and Society
- Derginin Tarandığı İndeksler: Science Citation Index Expanded (SCI-EXPANDED), Scopus, Compendex, Geobase, INSPEC
- Anahtar Kelimeler: Causal machine learning, Land value capture, Spatial policy simulation, Sustainable urban finance, Unearned urban rent
- Yıldız Teknik Üniversitesi Adresli: Evet
Özet
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. Because complex urban market dynamics make it difficult to isolate the causal impact of public investments, massive uncaptured rent fuels private speculation and undermines local government financing. A fundamental sequencing paradox compounds this: the value increment cannot be quantified before the investment, yet fiscal instruments require prior calibration. To address this chronic measurement problem, this study develops the LVC Oriented Investment Location Simulation (LOIS) model. This spatial decision framework estimates parcel level heterogeneous treatment effects (CATE) of public facility accessibility on land values using Double Machine Learning (DML) and Causal Forest algorithms, translating these into strategic site selection recommendations via AHP and TOPSIS multicriteria optimisation. Tested across 3562 parcels in Istanbul, LightGBM substantially outperformed traditional hedonic baselines in land value estimation (R²=0.671). A placebo test confirmed statistically significant CATE heterogeneity for green spaces, healthcare, and education. Facility specific spatial decay analysis revealed that park effects attenuate sharply within 800 metres (α=0.411), while educational facilities exhibit near zero distance sensitivity due to market saturation. Scenario perturbations confirmed optimal site selections are robust to institutional variation. Consequently, results demonstrate that unidimensional revenue maximisation exacerbates spatial inequality, whereas the LOIS framework enables municipalities to convert public investment planning into a self-financing instrument that simultaneously captures unearned rent and advances spatial justice.