Projections of fertility mortality and the population of the United States: 1990-2050.

The US Bureau of the Census in 1992 provided new projections based on cohort component methods of the population for 2050 which disaggregated population by 4 racial groups and allowed the subgroup rates to vary over the projection period. In this paper the figures are generated to approximate the new projections through the use of time series based projections and a constant for past mortality and fertility age specific patterns with a small number of parameters. Alternative scenarios are generated to demonstrate the effects of alternative assumption. The forecasts in this paper may be used as benchmark forecasts against which the official Census and other forecasts may be compared. This methodology also produces information on confidence levels appropriate for high and low variants of official forecasts. The methodology employs cohort component methods time series methods and an accounting for socioeconomic determinants of mortality and fertility. The utility of different forecasting methods is summarized. A full description of the time series method is provided. Essentially the projection methodology follows methods used by Rogers (1986) where age profiles of mortality fertility and migration are represented by variants of the multiexponential model. The models accuracy was checked by projecting 1985 and 1990 population from 1989 figures. The comparisons showed that there was some overestimation but in general the forecasts were reasonable. The principal results were that 1) population will reach 343 million by 2050 when constant rates are 2.01 for total fertility rate (TFR) immigration of 810753 and life expectancy from birth of 78.67 for females and 71.97 for males. 2) With a scenario of declining mortality and TFR of 2.25 population is expected to be 332 million by 2020 and 409 million by 2050. With high life expectancy population would b 337 million by 2020 and coupled with high fertility the population would be 347 million by 2020. With increases in all parameters population would be 330 million by 2020. Compared with the Census projections the forecasts are similar although the assumptions are different. Compared with the Ahlburg-Vaupel projections their lower fertility yields higher population in 2020 which is considered too high; differences occur due to higher longevity and net immigration figures.