Developing Male Fertility Forecasts to Inform Kinship Forecasts

Joanne Ellison , University of Southampton
Jason Hilton, University of Southampton
Jakub Bijak, University of Southampton
Erengul Dodd, University of Southampton
Peter W. F. Smith, University of Southampton

The ability to forecast the ways in which kinship networks are likely to evolve in the future provides a rich framework that can be used to answer many policy-relevant questions. Examples include the availability of informal care through the kin network, and the impact of different aspects of demographic change on social security. Kinship forecasts require projections of demographic rates for both sexes, and while there is an established literature on mortality forecasting methods for both males and females, fertility forecasting has thus far focused almost exclusively on females. Therefore, in this paper we develop a male fertility forecasting model, working within a Bayesian framework to appropriately quantify uncertainty. Preliminary work has focused on modelling male period fertility rates for England & Wales and the USA. We have used Bayesian P-splines to fit a smooth age-period surface to the observed rates that can then be extrapolated to future years. To reflect our prior knowledge that aggregate measures such as the total fertility rate change reasonably slowly over time, we investigate the application of a constraint to this measure. To assess the predictive accuracy of our proposed model, we compare our forecasts with some of the current best-performing models in the female fertility forecasting literature. We also investigate the joint forecasting of male and female fertility, borrowing strength across both sexes. Our findings have the potential to lead to more accurate sex-specific kinship projections, thus improving the reliability of the evidence for care policy that such projections can provide.

See extended abstract

 Presented in Session 104. Fertility and Reproduction: Bringing Men Back In