A wide range of methods have been employed in this report, as they are tailored to the various research questions and the nature of the data. For questions relating to local taxation and its effects on property prices and rents, the report uses panel data from French local authorities. Using a fixed-effects estimator, the report examines the correlation between each municipality’s level of exposure to the reform and the variables of interest.
For the final section on residential mobility, the methods used are more akin to differences-in-differences and discontinuity regressions. The individual-level data do indeed allow for the use of the thresholds for the Residential/local housing tax.
Data
Local taxation
The data sources used in this report are the local direct taxation assessment records (REI) from 2012 to 2022. This dataset, produced by the Directorate-General for Public Finances (DGFiP), records local tax data for the main direct local taxes and
at various levels (municipality, associations and similar bodies, inter-municipal bodies, department, region).
For total municipal revenue, the authors use the consolidated municipal accounts provided by the DGFiP and processed by the OFGL (Observatory of Local Finance and Public Management).
For property prices
The source is the Property Valuation Requests (DVF) database. The authors use the version provided by Etalab, which has standardised the codes, carried out geolocation and, most importantly, created a transfer identifier. The database is organised at the property level (meaning there may be several properties per transaction), but certain variables, such as the value of the transfer, are recorded at the transfer level.
For rents
The data used for this analysis are taken from the work by Chapelle and Eyméoud (2022). This data was collected from property listing websites (SeLoger.com and leboncoin.fr). It therefore reflects new rental listings, rather than the current stock of rental properties in France. This data source also focuses exclusively on the private rental sector. The advantage of this data (apart from being the only data available covering such a large part of France and spanning a significant time period) is that, by focusing on new listings, it represents the most responsive segment of the rental market. However, the drawback is that they are not representative of rents paid by the French population in general, but only of new rents. Finally, these data relate to proposed rents, which are not necessarily the rents actually agreed upon in tenancy agreements.
Mobility
The authors draw on three data sources.
The first database is INSEE’s Tax and Social Income Survey (ERFS). The ERFS contains both survey data and tax records on household income and taxes paid. This source therefore has the advantage of including households’ reference taxable income (RFR) as well as the Residential/local housing tax actually paid by the tax household. This information thus makes it possible to observe the actual discontinuity in income resulting from the Residential/local housing tax reform.
The second database used is the Survey on Resources and Living Conditions (SRCV). This source also comprises survey data and tax data.
The final data source is Fidéli (Demographic Files on Housing and Individuals). Produced by INSEE, Fidéli is a tax-based database providing a comprehensive representation of the French population. Fidéli is derived from income tax return records, Residential/local housing tax records, land registry records, and the Filosofi database (which provides data on disposable income at the tax household level). This database contains information on dwellings (physical characteristics such as the number of rooms, floor area and fittings; geolocation; the cadastral rental value used for local taxes, etc.) and the individuals occupying them (socio-demographic characteristics and taxable income).
Fidéli is linked to the Félin database (Sampled Income Tax Register). The linking technique used here is employed by the TAXIPP static Microsimulation tool, developed by the IPP research teams.
