Graph Neural Network Modeling for Islamic Wealth Distribution Based on Nidzomul Iqtishodi Principles

Abstract

Persistent inequality and poverty reveal structural flaws in conventional wealth distribution systems. Within Islamic economic thought, especially Nidzomul Iqtishodi fil Islam, economic problems stem not from scarcity, but from inequitable wealth distribution that prevents individuals from fully meeting their basic needs. This study aims to develop a predictive framework for Islamic wealth distribution by integrating these distributive principles into a modern deep learning model. This research employs a quantitative, applied computational approach using systems-based predictive modeling. Data are collected from publicly available macroeconomic and institutional sources, including Indonesia’s national Gini ratio (0.375), poverty rate (8.47%), sharia banking assets (IDR 980.30 trillion), and total Islamic finance assets (IDR 2,582.25 trillion). A Graph Neural Network (GNN) is utilized as the primary data analysis tool to construct a multi-layer socio-financial graph linking regions, zakat institutions, sharia financial intermediaries, and government transfers.The results show that a simulated increase in zakat-channeled transfers by 0.30% of GDP is associated with a predicted reduction in poverty of 0.18 ± 0.06 percentage points and a decrease in inequality of 0.003 ± 0.001 in the national Gini ratio. The proposed SDEI shows that the GNN-model improves out-of-sample poverty prediction accuracy by 11–19% compared to non-graph models