The uncertainty of risk preferences – a case study of storms and floods in Germany

This paper examines the association between natural disasters and risk preferences, a topic where existing evidence remains contradictory. We explore whether the selected method of data analysis is a potential cause of contradictory study results, comparing the frequently used two‑way fixed effects estimator with the recently introduced dynamic Difference-in-Difference estimator. Using data on natural disasters in Germany from 2008 to 2019, we find positive associations but also differences across the methods, including statistical significance and magnitude of the association. Using the dynamic Difference-in-Difference estimator helps mitigate the problem of heterogeneous treatment effects; however, we face considerable estimation uncertainty, which limits our findings and highlights the need for further research.