Time series modelling of birth data

Abstract

Three basic methods namely cohort component projection methods, statistical time series methods and structural modelling methods are discussed for the purpose of forecasting births, with the main focus on univariate time series methods. A general autoregressive integrated moving average model for birth time series is developed from the mathematical demographic renewal equation for births. The four-stage Box-Jenkins modelling method of model identification, estimation, diagnosis and forecasting is investigated in detail. This method is employed to model and forecast Australian birth time series. Finally, the comparison between time series forecasts and cohort component projections of births for Australia is made.