ABSTRACT
Policymakers often make decisions based on GDP, unemployment rate, industrial output, etc. The primary methods to obtain or estimate such information are resource-intensive. In order to make timely and well-informed decisions, it is imperative to come up with proxies for these parameters, which can be sampled quickly and efficiently, especially during disruptive events like the COVID-19 pandemic. We explore the use of remotely sensed data for this task. The data has become cheaper to collect than surveys and can be available in real-time. In this work, we present Regional GDP-NightLight (ReGNL), a neural network trained to predict GDP given the nightlights data and geographical coordinates. Taking the case of 50 US states, we find that ReGNL is disruption-agnostic and can predict the GDP for both normal years (2019) and years with a disruptive event (2020). ReGNL outperforms time-series ARIMA methods for prediction, even during the pandemic. © 2022 ACM.