Abstract
The article estimates flood risk perceptions by exploiting the different release dates of flood risk information around Paris from 2003 to 2012. This period is characterised by the absence of significant floods since 1955, making flood risk less salient. We apply a stacked event study to detailed property transaction data combined with geo-localised amenities. The results show that transaction prices for similar properties are 3-7% lower following the release of information if they are located in a flood risk zone, and that the effect persists, at least over the period we analyse. The results are robust to varying the control group to a neighbourhood at different distances from the flood risk boundary. The effect is more negative for flats on the ground floor. We find no evidence of sorting among buyers along different characteristics, in particular based on past exposure to flooding in their previous municipality. The results indicate a significant effect of flood risk information in a context where we can isolate it from the financial consequences of insurance cover and from flood damage per se.