UNCERTAIN FUTURES: STOCHASTIC FORECASTS/DECISION MAKING

点击翻译按钮获取中文摘要

The long-term goals of this project are to extend, develop, and apply methods for making stochastic demographic forecasts and for using these forecasts in decision making and socioeconomic analysis. The work proposed here has the following specific goals. 1. Extensions of the stochastic method of Lee and Tuljapurkar to probability distributions of population functionals; projections of quantities such as the social security tax burden and transfers; and sensitivity analyses. 2. Alternative models for vital rates: multiple principal components to capture complexity of mortality change; comparison of the forecasting performance of the Lee-Carter and McNown-Rogers models in terms of population forecasts; use of improved estimation procedures for parameters and standard errors; state-space techniques to capture long-term trends and/or cycles; exploration of models of fertility that incorporate Easterlin and Lee feedbacks; use of techniques of nonlinear time-series analysis to look for evidence of nonlinearity in fertility time series; estimation of nonlinear models so discovered using nonlinear versions of the Kalman filter. 3. Analyses of aggregation by age: multigenerational decomposition of uncertainty due to rates vis-a-vis population structure for growth rate and for population functionals; the relationship of such decompositions to ex post evaluations of forecast accuracy. 4. Methods for decision making in the presence of demographic forecast uncertainties, examining: (1) optimality and risk aversion in the context of different objective functions and (2) the dynamics of the OASDI trust fund considering alternative measures of risk.