ABSTRACT Household projections, traditionally focused on household headship, now require new methods to better capture changes in household structure due to the rise of single‐person households and the diversification of household types and sizes. Accurate projections for detailed household categories are essential for economic, policy, and environmental purposes but are challenging due to the low composition rates in many household types. This study proposes a new household projection method for Korea that addresses these challenges. We introduce a headship‐based household composition ratio and define marginal, conditional, and joint household composition ratios using a two‐way contingency table of household size and type. A sequential projection method combined with the cumulative rolling N+point modified exponential model is proposed to project these ratios. Projection results demonstrate that the new sequential cumulative rolling N+point modified exponential model can accurately and stably project household composition ratios, including categories with very small values, in both the short and long term.
Jeon et al. (Mon,) studied this question.