|
Tuli Amutenya , University of Southampton
Andrew J. Tatem, University of Southampton
Nikos Tzavidis, University of Southampton
Jessica Steele, University of Southampton
Duygu Cihan, University of Southampton
Population counts at subnational and local areas are vital for informed decision-making, service delivery, resource allocation and among many other applications. However, the population and housing census, the primary data source for population numbers, are typically collected decennially, less frequently and often delayed in these settings. To illustrate, Namibia postponed its 2020 census round three years before commencing in September 2023. Moreover, the traditional method employed by Statistic Offices to generate intercensal population projections often performs inadequately at small area scales due to data constraints. Detailed population data at small areas are often limited and become quickly outdated due to inherent demographic changes that take place at this scale, hence making it hard to forecast with traditional methods. This study explores alternative data sources and methods to enhance small area population projections, using Namibia as a case study. The study refines the cohort component method by incorporating post-baseline survey data. Additionally, geospatial data is disaggregated using top-down methods, and mobile phone data is used to update migration rates. The accuracy of these adjustments is validated against data from the 2020 Census mapping exercise and the 2023 census. Preliminary results suggest that the cohort component method is less accurate in areas with rapid population changes and urban regions. The study offers a unique opportunity to improve population projections by making use of available data, especially in developing nations with infrequent census undertakings and underdeveloped national registry systems. Keyword: small area population projections, cohort component method, official statistics, geospatial and mobile data
Presented in Session 52. Modelling Subnational and Spatial Variation