Stochastic methods in population forecasting

Nonfinancial Defined Contribution Pension Schemes in a Changing Pension World

“…In a study of Dutch forecasts, Keilman (1990) found that mortality forecasts systematically predicted too small numbers of the elderly. Similar findings from other countries have been reported by, among others, Alho (1990); Bongaarts and Bulatao (2000); and Keilman, Cruijsen, and Alho (2008). To see how the underestimation comes about, consider two historical examples that were chosen here because the forecast reports far surpass most other efforts in clarity and meticulous attention to detail.…”

“…Yet a historical look shows that forecasts have systematically overestimated mortality and underestimated life expectancy (e.g., Alho 1990;. Here the nature and causes of uncertainty in mortality developments are examined, and the statistical approaches commonly used by demographers and actuaries to derive projections of future life expectancy are discussed.…”

“…But the mortality forecasts of Wicksell (1926) and Whelpton, Eldridge, and Siegel (1947) turned out to be too high, because (a) the relative declines in the mortality of those of young and working ages have continued and (b) the rates of decline in older ages have increased. In fact, the conclusion to be drawn is that simply allowing past declines to continue would have made mortality forecasts more accurate (e.g., Alho 1990). This observation speaks in favor of trend-based statistical modeling.…”

Point and interval forecasts of age-specific life expectancies

“…In recent years, many authors have provided interval forecasts to measure the uncertainty associated with the point forecasts; see for example, Lutz and Scherbov (1998) for Austria, Alho (1998) for Finland, Keilman and Pham (2000) for Norway, and Tayman, Smith, and Lin (2007) for the United States. These methods have been motivated by earlier work on stochastic forecasts, by for instance, Lee (1974), Stoto (1983), Alho and Spencer (1985), Alho (1990) and Lee and Tuljapurkar (1994). The current paper is a contribution to the literature in this area.…”

Bayesian Methods in International Migration Forecasting

“…Accommodated within a stochastic approach, which quantifies uncertainty in terms of probabilities of future events. Keilman (2001), and Wilson and Bell (2004) distinguish three types of stochastic forecasts: extrapolation of time series (de Beer, 1990;Lee and Tuljapurkar, 1994;Keilman et al, 2001), propagation of historical forecast errors (Keyfitz, 1981;Stoto, 1983;Alho and Spencer, 1985;Alho, 1990;NRC, 2000), and probabilistic projections based on the expert judgement (Lutz et al, 1996(Lutz et al, , 1998.…”