N-point modified exponential model for household projections in Korea using multi-point register-based census data

Accurate household projections are essential for sectors such as housing supply and tax policy planning, given the rapid social changes like declining birthrates, an aging population, and a rise in single-person households that impact household size and type.Korea introduced its first register-based census in 2015, transitioning from fiveyear general survey-based approach to an annual administrative data-based census.This change in census allows for more frequent and effective capturing the rapid demographic shifts and trends.However, this change in census has caused challenges in future projection by the existing household projection model due to the rapid dynamics.This paper proposes a new household projection method, the N-point Modified Exponential Model (MEM), that accurately reflects register-based census data and mitigates the impact of rapid demographic changes, in three types: the Weighted N-point MEM, the Regression-based N-point MEM, and the Rolling Weighted N+point MEM.Using register-based census data from 2016 to 2020 to forecast household headship rates by age, household size, and household type to 2051, the N-point modified exponential model outperformed the existing model in both long-and short-term forecast accuracy, suggesting its suitability as a future household projection model for Korea.