In the French-speaking Belgian region of Wallonia, one third of households rent their accommodations. Three thirds of them represent private rental sector. Taking into account the important size of this sector and recognizing the weakness of the existing state of data collection, the administration of the Walloon Region has recently decided to create a system of residential rent observation. The detailed and updated information about housing attributes and rents should be used in housing policy revision, including more transparent rent estimation in private rental sector. The Center for Sustainable Housing Studies is responsible for the statistical and methodological management of this system.

The first wave of the observation system includes 2,500 face-to-face interviews among the private renters in the arrondissements of Nivelles and Charleroi. While the former represents to large degree the rich suburbs of Brussels, the latter is a mainly urbanized area well-known by its chronical problems of poverty, high crime rate and low quality of housing. In both arrondissements, it was asked 3.6% of households renting their accommodation in private sector._ _Though in hedonic regression literature the analysis of housing rents is rather rare, it is applied more and more often in recent years. In our study, a hedonic model of housing rents is constructed with internal and geographical attributes as explanatory variables. Neighborhoods are included in the equation as dummies. The aim of the study is to find which mode of data treatment provides better model performance and out-of-sample prediction: an overall model with all observations or market segments at the spatial level of the contrasting arrondissements (or at a lower spatial level). Spatial and geographically weighted regression methodologies are applied.