Agglomeration economies in the formal and informal sectors : a Bayesian spatial approach

This paper examines whether localized clusters of similar industries produce agglomeration economies in the formal and informal sectors. We develop a Bayesian method to estimate a spatial autoregressive model with an endogenous independent variable. We use establishment-level census data that cover both formal registered and informal unregistered establishments in Cambodia. We find that the density of local employment has a significantly positive effect on productivity in the informal sector, but little effect in the formal sector. For manufacturing, a doubling of employment density increases productivity in the informal sector by 9% through local linkages and by 17% through spatial multiplier linkages, leading to a 26% increase in total. A spatial network magnifies the local impact of agglomeration economies in the informal sector.