This paper introduces a novel methodology for computing local housing price indices at a national scale. The proposed approach addresses the limitations of traditional aggregated indices by capturing geographic heterogeneity in housing price dynamics.
The methodology relies on the application of a heuristic for the max-p-regions problem, a regionalization process that stratifies a country’s housing market into geographically consistent clusters, followed by the application of hedonic regressions made robust to noisy and sparse data. Applied to French housing data from 2010 to 2019, this method produces 876 consistent submarkets and their corresponding indices.
Our findings demonstrate the method's ability to reveal spatial patterns often obscured by national indices. While it captures trends at an aggregated level similar to official statistics, it also uncovers a much more contrasted trajectory of price changes across markets.
The methodology developed in this paper enables the construction of the most precise index database in France. A highly accurate index is created for each part of the country. This database allows for an analysis of housing price evolutions in the French market that has never been done before. This analysis reveals much more nuanced price developments than those described by existing benchmark indices because of the lack of localized data.
