Abstract:
When making decisions involving response and preparation for hurricanes and other disasters, individuals must determine how much risk they are willing to accept, and how much risk they believe they will face. There are many factors that can drive perceptions of risk. Notable drivers of risk perception are recency bias, and the hot-hand fallacy. Using the results of past hurricane forecasts and sales of emergency supplies, I find that responses to hurricane forecasts are largely influenced by past both recency bias and the hot-hand fallacy. Panic buying is significantly larger in counties with inaccurate most recent forecasts, and panic buying continues to get marginally worse as inaccurate forecasts are repeated.
Count of forecasted hurricanes that ultimately did not make landfall from 2008-2019.
Lightest counties had from 0 to 11 false forecasts while darkest counties had 36 to 46 false forecasts.
When consumers first learn of an impending natural disaster, they may be faced with substantial uncertainty regarding their future consumption and demand. As the exact location and severity of the event comes into focus, this uncertainty lessens and those who will be directly impacted tend to purchase emergency supplies all at once. This is known as “panic buying” and often leads to increased burdens on supply chains and market failures. Because this behavior is observed even in areas that are highly exposed to natural disasters, I use a dynamic consumer choice model to structurally estimate how the sudden announcement and severity of a natural disaster influences the consumer's decision to purchase and store emergency supplies. I find the costs of storing emergency supplies for future disasters is significantly higher than when estimated with traditional models helping explain the decision to wait.
Count of hurricane landfalls from 2008 - 2016.
Purple = 0, Blue = 1 or 2, Green = 3 to 5, Yellow = 6 to 10.
Natural disasters have become more frequent and severe causing some people to experience them multiple times in quick succession. This repetition of events increases the need to prepare for the possibility of natural disasters while also potentially changing preparatory behaviors through channels of learning and increased salience. I use an event study framework and heterogeneous regression analysis of counties in the eastern United States that were affected by the Atlantic hurricane season from 2008 - 2019 to better understand these channels. I find counties that faced few hurricane landfalls have statistically significant increases in sales of emergency supplies during the weeks surrounding any given landfall. My results also indicate that as counties experience more landfalls, the average sales of emergency supplies significantly decline. As the years between subsequent landfalls in a county increase, emergency supply sales weakly increase.
Event study of Battery Revenue per Capita