Although housing values in France have increased unevenly over the past twenty years, the spatial distribution of housing wealth has not been thoroughly studied. This lack of empirical research can be attributed to the absence of geocoded individual data, unlike countries such as Sweden or Norway. While French tax property files are available, they contain a structural error that makes it difficult to identify homeowners. Currently, these files enable the identification of housing properties detained at the department level, which is a major limitation when studying housing portfolios. To overcome this issue, we have constructed an invariant identifier at the country level using the civil status. Based on our error calculations, we estimate that the error correction process is highly satisfactory. It improves the measurement of the number of housing properties detained by each homeowner in France over the last decade. Furthermore, it enables the calculation of individual portfolio trajectories. By utilizing this enhancement of property tax records, we demonstrate the spatial aspect of property composition. We emphasize that the composition of properties is dependent on the level of house prices. The proportion of properties owned by multiple individuals increases significantly with the price level, while the proportion owned by single individuals decreases. Lastly, we reveal significant changes in the composition of properties over the past decade. The Société Civile Immobilière (SCI) has increased its market share independently of the market price level. In contrast, the market share held by single and multiple owners has decreased. Although the mechanisms behind this trend remain unclear, it raises new questions about potential consequences for the housing market.