Global panel analysis reveals key socioeconomic drivers of changing urban population density across countries, suggesting total urban land area could triple by 2100.
Urban expansion is strongly determined by changes in urban population and the density in which new residents settle. Consequently, assessments of future changes in urban area depend on population and density projections. While the former are well-established, the latter have received much less research attention. To fill this void, we set up a time series representing most countries around the globe describing urban population density and several explanatory variables. Using a panel regression approach with country-fixed effects we explain the variation in density levels over time. The results indicate that urban density increases with growing total urban population and decreases with increasing levels of urbanisation. The latter effect diminishes, however, when the urban population fraction increases even further. These results indicate a sequence of increasing, decreasing and stabilising urban densities with increasing urban population shares. This succession can be linked to subsequent development stages of urbanisation, suburbanisation and eventually reurbanisation. The latter stage remains rare in our time series, however, as only a few countries reach the corresponding high urbanisation rates. Furthermore, we find that urban densities decrease with growing national income per capita. This may indicate that higher incomes enable suburbanisation to more dispersed urban areas with larger residences but can also be associated with the advent of smaller households and the growth of non-residential sectors such as industry and commerce. Conversely, urban density increases with higher agricultural land values that we proxy with a novel primary production intensity variable. We use our explanatory results to determine country-specific future urban areas following a commonly applied socio-economic scenario (SSP2) and suggest the total global urban area may triple from 2015 to 2100.
No takes yet. Share an insight, caveat, or question.
Wielen et al. (2026) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: