The Not-so Marginal Value of Weather Warning Systems

- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner
Wednesday, 7 January 2015: 4:45 PM
221A-C (Phoenix Convention Center - West and North Buildings)
Benjamin M. Miller, Univeristy of California San Diego, La Jolla, CA

To date, most research regarding the effectiveness of weather warning systems has been limited to case studies. While such studies are particularly informative regarding how warning systems are applied in practice, it remains difficult to estimate the causal value of warning systems because there is no clear counter-factual of how individuals would have fared without them. Yet quantitative estimates of such marginal benefits are important for determining the optimal levels of investment in various warning systems.

To accomplish this task, this paper uses data from all recorded U.S. tornadoes between 1950 and 2012 to estimate the marginal benefit of National Weather Service (NWS) transmitters. After controlling for differences in tornado and location characteristics, variation in the installation dates of NWS transmitters is used to estimate the marginal value of transmitters in terms of prevented injuries and fatalities. Having a transmitter which broadcasts over a given country causally reduces tornado deaths and injuries in that county by 15 to 33 percent. Robustness checks confirm intuition that duplicative investment in transmitters does not provide additional protection, nor do transmitters reduce property damage.