Estimation of Future Vacancy Rates Throughout Japan Using Government Statistics

Abstract

The rise of vacant housing is a critical problem facing developed nations like Japan, causing detrimental impacts on local communities and housing market imbalances.Despite countermeasures, repeated widespread field surveys to assess vacant properties remain impractical due to prohibitive costs and time requirements.Interviews with municipalities have revealed a pressing need for forecasting future vacant housing numbers to develop effective mitigation strategies.Previous estimates have largely focused on prefectures, but there is a growing demand for more granular, municipality-level projections.This study leverages open data from the Population Census and Housing and Land Survey to develop a LightGBM machine learning model that predicts poorly maintained vacant houses at the municipal level across Japan.By utilizing Population Census variables such as population, age, and housing structure as explanatory features, and the 'other residences' category from the Housing Survey as the target, the model achieves highly accurate vacancy rate predictions for each municipality 3, 8, 13, and 18 years ahead until 2038.Notably, it can estimate current and future rates even for small municipalities under 15,000 people, where Housing Survey data is unavailable.The vacancy projections are published on WebGIS, providing a valuable reference for stakeholders formulating vacant housing policies and strategies nationwide.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 2 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.