Assessing the Accuracy of ARIMA and Hamilton-Perry Models to Forecast Population at the Intra-urban Level in Greater Mexico City

Population forecasts at the intra-urban level are crucial for governments to implement spatial planning and for businesses to identify potential markets. However, population forecasts are often produced at coarse geographic scales, namely national and regional. In addition, state of the art small-area forecasting methods are scarce and adjusted to work on Global North countries due to the lack of demographic data in the Global South. We aim to explore the accuracy of ARIMA and Hamilton-Perry models to forecast population totals and populations by age-group at the intra-urban level for the municipalities of Greater Mexico City. We used WorldPop 2000-10 population estimates as input data and performed retrospective forecasts for 2011-20. Comparing our forecasts with actual population counts, we found that ARIMA and Hamilton-Perry models return median errors of 1.5% and 3.1%, respectively, when forecasting population totals in a 10-year horizon. Median errors reached 5.7% and 4.1% for population forecasts by age-group. The age-groups 0-14, 15-39 and 40-64 performed particularly well, with errors below 10% in most municipalities, while the age-group ≥65 displayed higher errors, typically over 20%. We found no linear relationship between forecast errors and population size, but small municipalities often presented larger errors. We conclude that ARIMA and Hamilton-Perry models produce accurate forecasts and require minimum input data, which is an advantage in developing countries with no demographic data. Yet, both models present limitations. They are powerful when population change is consistent over time but cannot predict unexpected changes in demographic components.