Beyond Local Borders: A Spatial Durbin Analysis of Labour Migration and Poverty Dynamics in 1,063 Mahallas of Samarkand

Abstract

Labour migration is a defining economic dynamic in Central Asia, yet the spatial pathways by which it interacts with local poverty remain poorly understood. Previous empirical studies using spatially blind estimators systematically ignore regional spillover effects and spatial dependence. This study addresses this gap by analysing the spatial clustering of poverty and the structural associations of international labour migration with poverty across 1,063 mahallas in Uzbekistan’s Samarkand region. Utilising 2025 administrative microdata from the “Mahallabay” registry, we employ a Spatial Durbin Model (SDM) with a group-interaction spatial weights matrix to accurately capture local institutional realities. Empirical decomposition reveals highly significant positive spatial dependence in poverty itself, indicating that vulnerable mahallas’ poverty levels are closely correlated with those of their neighbours. While non-spatial estimators suggest strong poverty-reducing associations, SDM decomposition reveals localised direct associations, and spatial spillovers exhibit competing directional pathways that neutralise each other, yielding a statistically non-significant total effect. Furthermore, quantile regression suggests that access to agricultural capital serves as a safety net in severely impoverished mahallas, though wide confidence intervals warrant cautious interpretation. Our findings highlight that non-spatial estimators fundamentally misrepresent local developmental dynamics by ignoring spatial feedback. Consequently, this study notes that labour migration-poverty dynamics involve complex spatial trade-offs that complicate broad policy interventions, acknowledging that official registries may exclude unregistered marginalised populations.