LAUSR

Abstract This paper analyses the factors associated with the private transport modal share in cities of different wealth. We use the 1995 UITP Millennium Cities dataset and smaller samples of matching cities in the 2012 UITP Mobility in Cities dataset. Segmented bivariate analysis using the 1995 data showed that the associations between the private transport share and various socio-economic, transport, and land use variables are mostly non-linear or moderated by third variables. K-means clustering of the same 1995 variables then revealed three distinct groups of cities. Cluster 1 contains cities in developing countries with low private transport share and poor provision for both private and public transport. Cluster 2 contains high-income cities with high private transport share, low population density, and better relative provision and quality of private transport. Cluster 3 also contains high-income cities but with a moderate private transport share, higher population density, and better relative provision and quality of public transport. The evolution of cities from 1995 to 2012 showed that, as cities grow in wealth, they either move from Cluster 1 to Cluster 2 or from Cluster 1 to Cluster 3. Regression analysis provided further confirmation that the private transport share is explained by the variables that define the three clusters. Overall, the paper provides information for cities in developing countries to formulate combinations of transport and land use policies that can contribute to a transition towards sustainable transport systems.