LAUSR

Abstract HIV epidemics in developing countries are difficult to measure because the symptoms of HIV infections are non-specific and diagnosis relies on laboratory tests. We find that the nightlight data used in Henderson et al. (2012), originally compiled for estimating GDP growth rates in developing countries, can predict HIV incidence across countries. With economic performance, urbanization, and population density held the same, there remains a significant and robust negative association between nightlight and HIV incidence. In contrast, nightlight shows no association with other common health problems in developing countries. We posit that nightlight reflects the availability of basic public goods that prevent diseases with low infectiousness. We also construct nightlight-augmented HIV incidence for 68 countries with the poorest data quality.