Global transmission of COVID-19 — A gravity model approach

This paper aims to describe the spatiotemporal transmission of COVID-19 and examine how various factors influence the global spread of COVID-19 using a modified gravity model. Log-linearizing the model, we run a negative binomial regression with observational data from 22 January 2020 to 31 December 2020. In the first model, population size and GDP per capita are positively related to the sum of newly confirmed COVID-19 cases within a 10-day window; the values for both variables are statistically significant throughout the study period. However, the significance of geographic distance varies. When a single geographic source exits in the early stage, the value is statistically significant. In the intermediate stage, when disease transmission is explosive between countries, the distance loses its statistical significance due to the emergence of multiple geographic transmission sources. In the containment stage, when the spread of disease is more likely to occur within a country, distance becomes statistically significant. According to the second model, the government’s internal movement control and nonpharmaceutical intervention policy, percentage of the population over 70 years old, and population-weighted density are statistically significant and are positively related to the incidence of COVID-19. By contrast, average monthly temperature, international travel restriction policies, and political regimes are statistically significant and negatively associated with the dependent variable.