Regional Probabilistic Fertility Forecasting by Modeling between-Country Correlations

Bailey Fosdick, University of Washington

The United Nations (UN) Population Division is considering producing probabilistic projections of total fertility rate (TFR) using the Bayesian hierarchical model of Alkema et al. (2011), which produces TFR predictive distributions for individual countries. Obtaining probabilistic projections for country aggregates requires joint probabilistic projections of country-specific TFRs, taking account of the correlations between them. We propose an extension of the current model that estimates the correlation between country forecast errors as a linear function of time invariant covariates, namely whether the countries are contiguous, had a common colonizer after 1945, and are in the same UN region. This correlation model is incorporated into the current model's error distribution. TFR predictive distributions for 1990-2010 for each of the UN's primary regions show that the proportions of the observed values that fall within the prediction intervals from our method are closer to their nominal levels than those produced by the current model.