Estimating economic activity of South African provinces using satellite technology: An econometric panel data application

Abstract

This article studies the economic application of the remote sensing of night-time light emissions using satellite technologies within the South African context, focusing on its nine provinces as the unit of investigation. Observing human activity from outer space using satellite technology and remote sensing focusing on man-made light emissions presents real opportunity, especially where reliable national statistics are lacking. The methodology involves the estimation of panel data, autoregressive distributed lag and co-integration models that estimate provincial night-time light intensity as a proxy for provincial gross regional product (GDP). Both the ARDL and co-integrating models point to a long-run co-integrating relationship between GDP and night-time total light intensity, while the panel models suggest that the inverse Henderson elasticity for all provinces is 0.277 (cross-section fixed model). The ARDL models also establish a statistically significant short-term error correction term. The combined results suggest that a statistically significant long- and short-term relationship does indeed exist between provincial GDP and night-time lights within the South African and provincial context.