ABSTRACT
In this chapter we use Bayesian demographic accounts to study the growth of the New Zealand population. We begin with a national demographic system, with no age, sex, or any other way of classifying the population. We then extend the analysis to a regional system by adding a region dimension.
Data sources for the national system include census-based population estimates, administration-based population estimates, registered births, registered deaths, international arrivals, and international departures. The system models describe regularities in the true counts of population and events. Explicit data models are specified for the data sources. Estimates of population counts are most precise in census years, and least precise in years away from census years. We also examine estimates of the coverage ratios, and conduct sensitivity tests.
For the regional system, each existing data source is disaggregated by region. Furthermore, there are two additional data sources for internal migration: census-based transitions over a 5-year period, and administration based annual address change. The estimated number of internal migrations is generally less than the number of address changes, implying that the address change data overstate actual migration. The coverage ratios for the address change data reinforce this point.
