Abhatoo - Centre Nationale de Documentation

In view of the massive increases in urban population expected to take place in developing countries in the coming decades, demographers will need to refine and critically assess their methods for estimating and forecasting city populations. This paper draws upon a newly assembled and comprehensive cities database for developing countries, combining time-series information from the UN Population Division and spatial information from the Global Urban-Rural Mapping Project of Columbia University, to develop probabilistic forecasts of city growth. The forecasts make use of urban total fertility rates, child mortality, age structure, and migration, all of these being estimated from WFS, DHS, MICS, and other demographic surveys. The city growth models include both classical and Bayesian econometric models for panel data, and also include spatial econometric specifications incorporating linkages across cities.