Demographic Forecasting
Cover -- Title Page -- Copyright Page -- Contents -- List of Figures -- List of Tables -- Preface -- Acknowledgments -- 1 Qualitative Overview -- 1.1 Introduction -- 1.2 Forecasting Mortality -- 1.2.1 The Data -- 1.2.2 The Patterns -- 1.2.3 Scientific versus Optimistic Forecasting Goals -- 1.3 Statistical Modeling -- 1.4 Implications for the Bayesian Modeling Literature -- 1.5 Incorporating Area Studies in Cross-National Comparative Research -- 1.6 Summary -- Part I: Existing Methods for Forecasting Mortality -- 2 Methods without Covariates -- 2.1 Patterns in Mortality Age Profiles -- 2.2 A United Statistical Framework -- 2.3 Population Extrapolation Approaches -- 2.4 Parametric Approaches -- 2.5 A Nonparametric Approach: Principal Components -- 2.5.1 Introduction -- 2.5.2 Estimation -- 2.6 The Lee-Carter Approach -- 2.6.1 The Model -- 2.6.2 Estimation -- 2.6.3 Forecasting -- 2.6.4 Properties -- 2.7 Summary -- 3 Methods with Covariates -- 3.1 Equation-by-Equation Maximum Likelihood -- 3.1.1 Poisson Regression -- 3.1.2 Least Squares -- 3.1.3 Computing Forecasts -- 3.1.4 Summary Evaluation -- 3.2 Time-Series, Cross-Sectional Pooling -- 3.2.1 The Model -- 3.2.2 Postestimation Intercept Correction -- 3.2.3 Summary Evaluation -- 3.3 Partially Pooling Cross Sections via Disturbance Correlations -- 3.4 Cause-Specific Methods with Microlevel Information -- 3.4.1 Direct Decomposition Methods Modeling -- 3.4.2 Microsimulation Methods -- 3.4.3 Interpretation -- 3.5 Summary -- Part II: Statistical Modeling -- 4 The Model -- 4.1 Overview -- 4.2 Priors on Coefficients -- 4.3 Problems with Priors on Coeffcients -- 4.3.1 Little Direct Prior Knowledge Exists about Coefficients -- 4.3.2 Normalization Factors Cannot Be Estimated -- 4.3.3 We Know about the Dependent Variable, Not the Coefficients -- 4.3.4 Difficulties with Incomparable Covariates