This is an empirical study about the formation of residential prices in Paris. It is asked to which extent the formation of prices is guided by selected districts that influence the prices in other parts of the city. The analysis is performed with econometric models of the SAR-type. The spillover mechanism of prices is ranked according to the partial contribution to spatial autocorrelation of the districts. We use real estate data collected by the French notary bureau BIEN containing nearly 30 000 transactions, and  population data offered by the IRIS-database and the statistical bureau INSEE. In the analysis, we focus our attention to the "reflection" problem that concerns the identification of causal effects in the SAR-models. Doing so we can offer results that explain what drives the diffusion of a district, obtaining new insights into the leading role of population data which make the price diffusion generated by districts stronger or weaker.