Estimating socioeconomic variables in Bolivia using satellite-based nighttime light images and electricity consumption data

Abstract

Electricity consumption (EC) has a great potential to proxy socio-economic data because it is closely related to household income levels. Existing research recognizes the use of satellite-borne nighttime light (NTL) products as an effective method to estimate EC at different scales. Moreover, nighttime light (NTL) radiance has been shown to aid in the estimation of specific socio-economic indicators, like poverty, in areas with temporal or spatial data gaps. Nevertheless, little research has been done to explore this potential at sub-national level in Latin America, a region with particular urbanization and land cover patterns. This research examines (1) the relationship between NTL and EC using regression models, for Bolivian municipalities, and (2) the suitability of NTL and EC to estimate poverty at municipality level in Bolivia. First, we show that a positive correlation exists between the EC and NTL, in accord to similar research conducted in African or Asian countries. Second, we observe a moderate correlation between NTL and poverty rates, superior in strength to the relationship between EC and poverty. For this, we built simple and multiple linear models and we conducted change detection and time-series analysis to explore spatio-temporal patterns. Our results confirm that EO-derived products data can complement official data for local estimation, but that temporal estimations still lack precision.