Household and population projections at sub-national levels: an extended cohort-component approach

This paper describes the core methodological ideas, the required input data, and estimation issues of the extended cohort-component approach to simultaneously project household composition and population distributions at sub-national levels. We assess the projection accuracy of this approach by calculating projections from 1990 to 2000 and comparing projected with the census-observed counts in 2000 for the 50 states and the District of Columbia and for sets of randomly selected 25 counties and 25 cities which are more or less evenly distributed across the United States. The comparisons show that most absolute percent errors of the main indices of household and population projections and the corresponding census observations are small -less than three percent -and almost all errors are less than ten percent. We then report illustrative household projections from 2000 to 2050 for the 50 states and the District of Columbia, and household/housing projections for the small town of Chapel Hill, North Carolina up to 2015 in order to demonstrate the practical capabilities of the new approach. Among many interesting numerical outcomes, the aging of American households over the next few decades across all states and the aging of the housing market in Chapel Hill are particularly striking trends in the projections.