|
Abstract:
|
Many of the most important applications in demography, from population estimation to the analysis of labour markets or population health, are special cases of the general problem of estimating demographic accounts. A demographic account is a set of cross-tabulations of population and events, linked by fundamental accounting identities: for the whole population, and for every subpopulation, the population count at the end of each period equals the population count at the beginning, plus counts of entries, and minus counts of exits. Estimating demographic accounts is challenging because of the need to synthesize multiple incomplete and noisy datasets, and the need to obtain counts that conform to the accounting identities. We present a fully Bayesian framework for the estimation of demographic accounts that is flexible and transparent, and that provides comprehensive measures of uncertainty. Outputs from the framework include estimates of the true counts of people and events, estimates of the associated demographic rates, and measures of the quality of the data sources. We illustrate the framework with the example of estimating a demographic account for China.
|
