The billion-dollar disaster time series should provide us a lesson on how easy it is to promote a simple message that fits a particular narrative but which is completely misleading and even outright wrong.
The U.S. National Oceanic and Atmospheric Administration (NOAA) counts the number of disasters in the United States that result in losses of greater than $1 billion, starting in 1980. Over the past three decades that count has shown a sharp increase, from five or less such disasters each year in the decade of the 1980s to ten or more in each of the past 4 years.
That increase must be due to climate change, right? Actually, no. The billion-dollar disaster tally is easy to understand, simple to communicate, but—regrettably—incredibly misleading and just plain bad economics.
Before proceeding, it is important to underscore that climate change resulting from the emission of greenhouse gases from the burning of fossil fuels poses significant risks to our collective futures, including influences on extreme events. As a consequence, it makes sense to focus policy on the mitigation of emissions and adaptation to reduce vulnerability and exposure to weather and climate.Today In: Business
But the importance of climate change as a policy issue does not mean that the subject gets a free pass when it comes to scientific accuracy, and especially claims made by authoritative bodies like NOAA. To the contrary, building support for action on climate requires maintaining the highest standards of scientific integrity.
The billion-dollar disaster tally, even though it is popular, doesn’t meet that standard. The biggest problem with the time series is that it is based on a threshold of $1 billion, but both the value of the dollar changes over time (inflation) and, more importantly, the value of property and wealth subject to losses from extremes has grown dramatically over time.
When NOAA first released the billion-dollar disaster dataset in 2012, it neglected historical events that, after considering inflation, would have exceeded the billion-dollar threshold. For example, a disaster that caused $900 million in losses in 1980 (in 1980 dollars) would not have been included, although the actual loss amount would have exceeded $1 billion in contemporary times. After this was pointed out, NOAA corrected the oversight and added 19 additional events to their dataset, warning appropriately: “Caution should be used in interpreting any trends based on this graphic for a variety of reasons.”
But the inflation snafu revealed a far deeper problem with the use of the billion-dollar threshold: U.S. disaster take place at the intersection of a changing and variable climate and a nation growing in population, wealth and development. Consider that an identical hurricane making landfall in Texas in 1980 versus 2019 would result in vastly different loss totals, because today there are simply more people in more homes with more stuff than thirty years ago. In 2012, I identified nine disasters from 1980 not included in the NOAA tabulation that would likely have exceeded a billion-dollar threshold has those events occurred in 2011.
The NOAA billion-dollar disaster dataset is not a reliable indicator of trends in disasters or their costs, and it certainly is not a time series that says anything meaningful about changes in climate. Anyone wanting to look at trends in climate and weather, including extreme events, should always look first at data on climate and weather, not economic loss data.
So how might NOAA improve its economic loss record for U.S. disasters? The answer is simple: Do away with the billion-dollar threshold and look at the entire record of losses. Even better, address the effects of a growing U.S. economy, and greater loss potentials over time, by normalizing disaster loss data based on GDP (or other factors).
That is exactly what I did in the graph below, based on data from the U.S. Office of Management and Budget (GDP) and Arizona State University (hazard losses). The graph shows U.S. hazard losses from 1980 (the start of the NOAA dataset) to 2016 (the end of the ASU dataset). There is clearly no upwards trend. The apparent slight downward trend results from the “drought” of major hurricanes from 2006 to 2016, which has since ended in 2017 and 2018.
Compare that graph to NOAA’s billion-dollar disaster tally, shown below, over the same time period. You can easily see how the NOAA tally has potential to mislead.
To their credit, early on NOAA recognized that there were methodological issues in its approach to collecting and sharing disaster loss data, and commissioned a study of the dataset and methodology, which was peer-reviewed and published in 2013.