Top Lights - Bright Spots and their Contribution to Economic Development

Satellite data on nighttime luminosity is an increasingly popular proxy for economic activity in developing countries. However, their use for analyzing inequality and convergence on a global scale is severely limited by top-coding of the NOAA satellite images, which fail to accurately capture the brightness of large and densely populated cities, as well as by comparability problems between satellites. As a result, they severely underestimate differences between urban and rural regions, and developed and developing countries. We propose a new and easy-to-use procedure to correct for top-coding in nighttime lights, which borrows from the top incomes literature. We show that just as top incomes, top lights are Pareto distributed. We then derive simple formulas for the top-coding adjusted spatial Gini coeffcient and top-coding adjusted average light intensity. Using data for Germany we show that by correcting for top-coding of the top 2% of lights, we can account for up to 40% of the difference between saturated and unsaturated satellites. We also analyze corrections for between and within satellite measurement errors. Finally, we present three economic applications to determine where the influence of top-coding is most severe. We show that top-coding and satellite calibration aspects estimates of the income elasticity of light, regional inequalities and urban-rural differences.