Global evidence on productivity effects of road infrastructure incorporating spatial spillover effects

This paper empirically analyzes the productivity effects of road infrastructure incorporating spatial spillover effects using a global database, which was originally developed in 2010 at a grid-cell scale of 0.1-degree covering all land areas in the world by integrating national-level statistical socio-economic data with available global data sources of nighttime lights imagery, global population database, and global road network. A macroscopic Hicks-neutral Cobb-Douglas production function, containing input factors of capital stock, labor force, and road infrastructure is estimated with the global dataset of 14,565 aggregated grids at 1.0-degree grid-cell scale, using ordinary least squares, spatial error model, spatial lag model, and spatial Durbin model (SDM). The statistical tests suggest that there is significant spatial dependence particularly in the error term and the overall results of the model estimations indicate that the SDM outperforms other models. The estimates of SDM further show that the direct impact of road infrastructure is significantly negative, spatial spillover effect is significantly positive, and overall effect is positive but insignificant.