Reconciling forecasts of infant mortality rates at national and sub-national levels: Grouped time-series methods

摘要

Mortality rates are often disaggregated by different attributes, such as sex,state, education, religion or ethnicity. Forecasting mortality rates at thenational and sub-national levels plays an important role in making socialpolicies associated with the national and sub-national levels. However, baseforecasts at the sub-national levels may not add up to the forecasts at thenational level. To address this issue, we consider the problem of reconcilingmortality rate forecasts from the viewpoint of grouped time-series forecastingmethods (Hyndman et al., 2011). A bottom-up method and an optimal combinationmethod are applied to produce point forecasts of infant mortality rates thatare aggregated appropriately across the different levels of a hierarchy. Weextend these two methods by considering the reconciliation of intervalforecasts through a bootstrap procedure. Using the regional infant mortalityrates in Australia, we investigate the one-step-ahead to 20-step-ahead pointand interval forecast accuracies among the independent and these two groupedtime-series forecasting methods. The proposed methods are shown to be usefulfor reconciling point and interval forecasts of demographic rates at thenational and sub-national levels, and would be beneficial for government policydecisions regarding the allocations of current and future resources at both thenational and sub-national levels.