This paper studies financial service provision booked through offshore financial centers (OFCs). Based on several novel data sources and recent advances in event study methodology, I exploit the natural experiment of re-occurring hurricanes hitting small islands and compare local reactions to reactions in financial service activity. I find that local conditions, captured by monthly satellite data on nightlight intensity, deteriorate significantly for nine months. However, in OFCs, the international bank sector does not react. Non-OFC islands on the other hand do show strong negative reactions. Similar (non-)reactions are visible in equity prices. Additionally, a link of OFC service provision to activity in London, Tokyo, and New York is visible in leaked data. Finally, a long term relationship between offshore finance and local development is absent, but only on OFCs. These results indicate that international regulation attempts that aim at forcing OCFs to provide information on financial service activity could be targeted better, they show that we mis-allocate financial risk to OFCs, and they cast doubt on offshore finance as a valid development strategy.