Household and Population Projections at Sub-Nationals Levels: An Extended Cohort-Component Approach

Yi Zeng, Duke University and Peking University
Zhenglian Wang, Duke University
Danan Gu, Duke University

This paper describes the core methodological ideas and data/estimation issues of an extended cohort-component approach to simultaneously project household composition and population distributions at sub-national levels. We assess this approach by using projections from 1990 to 2000 and from 2000 to 2006 and comparing the projected with the census-observed estimates in 2000 and ACS-observed estimates in 2006 for 50 states and DC, and three sub-state areas. The comparisons show that most absolute errors between 1,475 pairs of indices of household and population projections and corresponding census and ACS observations are small – less than three percent – and almost all errors are less than ten percent. We then report illustrative household projections 2000-2050 for 50 states and DC, and household/housing projections for the small town of Chapel Hill 2000-2015. The aging of American households over future decades across all states and aging of housing market in Chapel Hill are particularly striking trends in the projections.