Floods do not sink prices, historical memory does: How flood risk impacts the Italian housing market

TL;DR

This paper tests the historical memory hypothesis for flood risk in the Italian housing market using a novel dataset of ~550,000 ISP mortgage transactions. It combines hedonic regressions with a difference-in-differences design and constructs a region-level flood-awareness measure from EM-DAT and ISPRA data to test how repeated flooding shapes price discounts. The main findings show no price penalty from single floods for at-risk-but-unhit homes, but a progressive discount in highly aware, frequently flooded regions up to about 4%, with high-income buyers driving the discounts and low-income buyers more likely to move to risk-prone areas. The work highlights the role of cultural and institutional factors in climate-risk pricing and has policy implications for risk disclosure, insurance incentives, and resilience investments.

Abstract

Do home prices incorporate flood risk in the immediate aftermath of specific flood events, or is it the repeated exposure over the years that plays a more significant role? We address this question through the first systematic study of the Italian housing market, which is an ideal case study because it is highly exposed to floods, though unevenly distributed across the national territory. Using a novel dataset containing about 550,000 mortgage-financed transactions between 2016 and 2024, as well as hedonic regressions and a difference-in-difference design, we find that: (i) specific floods do not decrease home prices in areas at risk; (ii) the repeated exposure to floods in flood-prone areas leads to a price decline, up to 4\% in the most frequently flooded regions; (iii) responses are heterogeneous by buyers' income and age. Young buyers (with limited exposure to prior floods) do not obtain any price reduction for settling in risky areas, while experienced buyers do. At the same time, buyers who settle in risky areas have lower incomes than buyers in safe areas in the most affected regions. Our results emphasize the importance of cultural and institutional factors in understanding how flood risk affects the housing market and socioeconomic outcomes.

Figures (15)

• Figure 1: Regional awareness trends computed on the EM-DAT data. The awareness is computed day-by-day according to Equation \ref{['eqn:awareness']} with $\tau=10$ years. The spikes are due to flood events.
• Figure 2: Graphical description of the regression framework. Panel a): An example to illustrate our identification strategy for homes at flood risk. We show two homes transacted in our dataset in yellow (after adding some noise to their positions to preserve privacy) in the village of Usella, Prato. We also show the boundaries of the census tract in white and the ISPRA risk areas in shades of blue (dark blue: high and medium risk, light blue: low risk). The home shown as a square at the top is not classified as at risk, while the other home shown as a circle is classified as at risk because it is close to the river. We argue that the residual difference in price between these two properties, after controlling for all other characteristics, is due to being located in the flood zone. Panel b): Graphical description of HitRisk and NoHitRisk homes considering the municipality of Forlì, one the cities affected by the Emilia-Romagna flood in 2023.
• Figure 3: Time evolution of the home price variation for the 2023 Emilia-Romagna flood. Referring to Equation \ref{['eq:diffindiff']}, the left panel reports the coefficients $\beta_1$ and $\beta_2$, the central and right panels show the interaction coefficients $\delta_1$ and $\delta_2$ up to 15 months after the flood of Emilia-Romagna 2023. The temporal bins can be interpreted as follows. The labels pre and post identify whether a transaction has been completed before or after the considered flood event. The label y indicates years and the label m indicates months. We used the bin pre 1y as a reference. Each point represents the average value of the regression coefficient and the error bars indicate the 95% confidence intervals. We used spatial fixed effects at the OMI microzone level. R$^2$ = 0.74, observations = 35623.
• Figure 4: Effect of being at flood risk on home price, disaggregated by region. Panel a): The dots show the point estimate of the coefficients, while the error bars show 95% confidence intervals. For graphical purposes, we limited the x-axis between -0.1 and 0.1 (although not fully shown, the average price variations for Molise and Basilicata are 0.19 and 0.15, respectively). Regions are sorted according to time-averaged awareness: the first region (Emilia-Romagna) shows the largest average awareness value while the last region shows the lowest value. Panel b): Map of the Italian regions colored according to the average awareness at time of sale ($\tau=10$ years, see Section \ref{['sec:frequency']}). Northern regions: Piemonte, Valle d'Aosta, Liguria, Lombardia, Trentino-Alto Adige, Veneto, Friuli-Venezia Giulia, and Emilia-Romagna. Center regions: Toscana, Umbria, Marche, Lazio, Abruzzo, and Molise. Southern and Islands regions: Campania, Puglia, Basilicata, Calabria, Sicilia, and Sardegna.
• Figure 5: Flood risk effect on home prices disaggregated across awareness, mortgage applicant age, and income class of the buyer, Equation \ref{['eq:quadruple']}. The left panel shows the home price variation due to flood risk for transactions financed by the mutuo giovani, and disaggregated by our classification of historical memory (i.e., awareness) and buyers' income class. The right panel reports the same analysis for transactions financed by a standard mortgage. The buyers' income class is estimated considering the top, middle and bottom terciles of the income distribution within the same region (see Section \ref{['subsec:profile']} for further details).