November 2023
Tanner Regan (George Washington University)
Giorgio Chiovelli (Universidad de Montevideo)
Stelios Michalopoulos (Brown University, CEPR and NBER)
Elias Papaioannou (London Business School, CEPR)
Abstract: Satellite images of nighttime lights are commonly used to proxy local economic conditions. Despite their popularity, there are concerns about how accurately they capture local development in low-income settings and different scales. We compile a yearly series of comparable nighttime lights for Africa from 1992 to 2020, considering key factors that affect accuracy and comparability over time: sensor quality, top coding, blooming, and, importantly, variations in satellite systems (DMPS and VIIRS) using an ensemble, machine learning, approach. The harmonized luminosity series outperforms the unadjusted series as a stronger predictor of local development, particularly over time and at higher spatial resolutions.
JEL Codes: O1, R1, E01, I32
Keywords: Night Lights, Economic Development, Measurement, Africa
