Mathematical statistician Nassim Taleb recently released a COVID-19 inspired paper looking at single point forecasts for fat tailed events.
Taleb is most famous for his book “The Black Swan” and his work in probability and uncertainty. This paper looks at using a single point estimate to measure risk for fat tailed events. An event with a fat tail is an event with a probability distribution of outcomes which skew to one side. Taleb essentially argues you can’t look at the average outcome for these type of events.
An event with a fat tail is an event which part of the range of outcomes stretches out but is significantly less likely than the body of the distribution. Pandemics, like COVID-19, can be thought of as fat tailed events. Think of it this way: most hypothetical pandemics death count estimates are around a single point– say 100,000 deaths– but the distribution curve stretches all the way to 2,000,000 deaths. While 2,000,000 deaths is unlikely, that outcome is possible.
COVID-19 is one of these fat tailed events. A disease outbreak causing the US– and the world– to shut down and killing over 100,00 Americans and heading to 200,000 (and maybe 300) is not a common outcome of a disease outbreak. It’s an outbreak from the tail of the distribution. It’s also an outcome that the US had the opportunity to prepare for but didn’t. President Bush tried to improve US preparedness for outbreaks, but Presidents Obama and Trump didn’t take the risk seriously and let our preparedness slide back. Bill Gates warned the world that we weren’t prepared for a global pandemic. Those warnings fell on deaf ears.
COVID-19 should be a lesson to take unlikely but potentially devastating events seriously. One of those events is climate change. The Intergovernmental Panel on Climate Change regularly produces a range of possible temperature changes from the increase in CO2.
I’ve often heard conservatives dismiss climate change because the likely outcome isn’t that bad. That’s the wrong way of thinking about the risk of climate change. While the point estimate forecast (mean outcome) of the temperature change is worth knowing, it is not the most relevant data point right now. The range of outcomes is significantly more important. Yes, a two or three degree Celsius increase in global temperatures may be multitudes more likely than a six or seven degree Celsius increase but that six or seven degree increase is a real possibility. While it may be unlikely, the result would be devastating and apocalyptic— and that massive cost is what makes it important to pay attention to.
We can take actions now to mitigate the risk and impact of climate change. We choose not to for global pandemics and now we are paying the price. The cost of pandemic preparedness will be dwarfed by the lives and economic output lost because we choose not to be proactive. We face the same choice with climate change, but with drastically wider reaching consequences– we can choose to mitigate the risk now or gamble the future of humanity and our planet. Remember, there will be no redo.
Exit quote: “Risk management (or policy making) focuses on tail properties not the body of probability distributions. For instance, Holland has a policy to calibrate their dams not on average height of sea levels but on the properties of the maxima -all it takes is one mistake to cause disaster.”
If you enjoy articles like this or our other content, we encourage you to subscribe below!