Being in Miami, I found this paper rather interesting. Richard Evans (BYU), Yingyao Hu (Johns Hopkins), and Zhong Zhao (IZA) test whether hurricanes can make you pregnant. Sort of. The idea is that a natural disaster might encourage people to stay put in doors, and with nothing else to do people may use that opportunity to make babies.
For all US counties on the Atlantic and Gulf coasts, they have data on tropical storm watches, tropical storm warnings, hurricane watches, and hurricane warnings. These are listed in order of their expected severity. Basically, if your community is under a tropical storm watch, its raining pretty heavily outside. If your community is under a hurricane warning, its not just wet, but you're also worried about your house blowing away. Compared to the case without any watches or warnings, the authors find that being under a tropical storm watch significantly increases the number of births in your community 9 months later (consistent with the idea that since it's misserable outside, we might as well keep warm inside). However, a hurricane warning significantly decreases the birth rate 9 months later. I guess that people are too worried about their house blowing for any romance to take place. (For the intermediate storm advisories: A tropical storm watch has a positive but insignificant effect on the birth rate, while a hurricane watch has a negative but insignificant effect.)
The paper is forthcoming in the Journal of Population Economics. Download it here.