Abstract

Since 1961, Population Census of Hong Kong, China takes place every ten years, supplemented by a sample based by-census between two full censuses. The data collection operation in each round was short, yet demanded substantial resources to plan and execute, and was subject to considerable operational risks due to its large scale.

As a major step to modernise the population censuses and by-censuses, and to reduce their operational risks and costs in the long run, Hong Kong is embarking on a re-engineering initiative on its Population Census starting from the 2026 round –
(a) conducting only a sample enumeration using the “long form” every five years, in lieu of a full census once every ten years and a sample based by-census in between;
(b) extending the data collection period from 1.5 months to one year; and
(c) utilising government administrative data to estimate total population size accurately and replace some census questions where appropriate.

By implementing these re-engineering measures, Hong Kong anticipates saving approximately 40% of the resources typically required in a full cycle comprising a full census and a by-census. Additionally, the use of government administrative data more extensively will contribute to enhanced data quality and reduced respondent burden. Pursuant to the re-engineered initiatives, new data collection approaches and estimation methodologies have been developed, such that Census statistics will still be referencing to the Census moment, i.e. mid-year.

This paper provides an overview of the re-engineering initiatives and the incidental changes in data collection approaches and estimation methodologies to support the re-engineered Census operation.