Trade in the spotlight: enhancing gravity model predictions with nightlights and population-weighted distance measures
Does a gravity model fit better when bilateral distance reflects where population and measured economic activity are located within countries?
- Status
- Preprint · SSRN 5202676
- Role
- Coauthor with Elizaveta Gonchar
- Authors
- Elizaveta Gonchar, Ian Helfrich
- Method
- gravity model · VIIRS nightlights · population-weighted geodesic distance
- What this record does not establish
- The preprint does not establish the planned panel, transport-network extension, or structural-gravity validation; those remain development work with no completed results.
Public artifactsPreprint
Abstract
Summary
The distribution of population across and within countries naturally relates to the distribution of economic production. We explore differences in gravity-model estimates of trade that take spatial factors of population distribution seriously, introducing a novel geodesic distance measure between countries using population-weighted centroids as endpoints for bilateral distance. Canonical gravity models have relied on time-invariant CEPII distance measures since 2004; we show that incorporating annual VIIRS nightlight intensity and gridded population yields measurably better gravity fit. The OVDL (Origin-VIIRS to Destination-LandScan) measure is the headline construction. April 2025 preprint; the full panel and structural-gravity validation are left to a companion effective-distance paper in preparation.
Methods and notes
April 2025 working paper. The OVDL measure introduced here is the seed of planned effective-distance panel work. That project is in preparation and proposes a multi-modal least-cost-path construction, structural-gravity validation, and applications to recent chokepoint shocks; no completed panel, counterfactual result, or submission target is claimed here.
BibTeX
Citation
@misc{gonchar2025trade,
author = {Gonchar, Elizaveta and Helfrich, Ian},
title = {Trade in the spotlight: enhancing gravity model predictions with nightlights and population-weighted distance measures},
year = {2025},
note = {Preprint, SSRN 5202676},
url = {https://ssrn.com/abstract=5202676}
}