Who knows better? Homesellers and homebuyers self-reported estimations

We use self-reported estimations collected on an Internet platform to study the formation of housing price beliefs of sellers and buyers of the Paris region. Comparing the users’ estimates with real prices of transactions, we are able to evaluate how accurate they are in predicting the selling price and to identify factors that influence their opinion. We confirm the already studied upward bias of homeowners but bring a first evidence that buyers are unbiased. Our results confirm that market participants follow an anchoring and adjustment scheme to set their opinion on housing price. Users' beliefs are influenced by the results of the platform Automated Valuation Model, which acts as an anchor. Buyers are less influence by this reference point, and the more the sellers progress in their sale project, the less their opinions rely on the model estimations. Finally, the estimates of owners whose property has lost value since they purchase it are anchored to its purchase price and less correlated to the AVM results. We interpret these results as proof of an updating mechanism of the buyers’ and sellers’ beliefs along their search for a counterpart.