Mapping Physical Climate Risk Exposure in Euro Area firms: Linking E-PRTR and RIAD with a Fuzzy Match

In this analysis, we develop a methodology to better capture the exposure of euro area firms to physical climate risks by incorporating the geographic distribution of their production facilities. We do so by linking two complementary datasets: the Register of Institutions and Affiliates Data (RIAD), which provides firm-level information, and the European Pollutant Release and Transfer Register (E-PRTR), which contains facility-level data. Because these datasets lack a common identifier, we apply a fuzzy string matching algorithm enhanced with machine learning to link firms to their facilities.

We replicate the methodology developed by the European System of Central Banks (ESCB) to estimate expected annual losses (EAL) from floods and windstorms. Our results show that relying solely on headquarter locations can significantly underestimate flood risk exposure, particularly for river floods. In contrast, windstorm risk appears less sensitive to the inclusion of local units. We also highlight regional differences in risk exposure across NUTS 3 regions, underscoring the importance of granular data in climate risk assessments.