Abstract
A cohort-component method is used to construct historical annual age/sex disaggregations of total population. In econometric models of six substate areas, regressions are estimated with and without the disaggregated data for five model variables expected to be dependent on the size of particular population subgroups. Statistical tests are performed to determine significant differences in residual variance of estimation with and without the disaggregated data. Although age/sex population projections may be a desired model output per se, the results indicate that the population subgroup data generally do not improve estimation of other model variables.

This publication has 6 references indexed in Scilit: