Risks are greater than we think
The recent horrific tragedy in Japan caused by a massive earthquake and the subsequent tsunami reminded me of the 2008 financial crisis and how markets were gyrating up and down five percent a day. What do the two events have in common? Both were deemed statistically likely to happen very infrequently, such as once in a hundred years or more.
In the depths of the financial crisis I recall debating with an investment manager whether the financial disaster was an 8 or 9 sigma “tail” or unexpected event. In a normal distribution of possible outcomes, each sigma from the mean (one standard deviation from the mean or most likely outcome) is less and less likely. As a normal distribution shows 3 sigma covers 98% of statistically possible outcomes so an 8 or 9 sigma event is extremely unlikely (much less than 2% probability) and the “tail” or outer most part of the distribution is very narrow or thin.
In the months and years since the financial crisis I have realized that debating whether an event is 8 or 9 sigma is not only pointless and academic, the reality is that it was probably more like a 2 or 3 sigma event and everyone’s financial models were simply wrong. In other words, the
Related to this I see three trends that, when combined, do not bode well for businesses or nations.
The first is that we seem to be having more and more tail events in all areas of life. Natural disasters are becoming more frequent and severe, financial shocks feel like they can happen any time, commodity prices are ever more volatile, and sovereign nations are becoming less predictable. This could be simply the perception that tail events have become more frequent given the availability of information across the globe in the internet age but it seems as though it is more than just perception.
The second trend is the continuously expanding reliance on computing power and systems to develop complicated statistical models that are supposed to help predict the future. I think we are already past the point where many financial models cannot be fully understood by the average person.
The third and perhaps most concerning trend is that businesses, governments, and individuals are focused on short term performance and find their balance sheets depleted resulting in little to no cushion to absorb the next unexpected catastrophe, whether a natural or financial disaster or otherwise.
As financial managers or
Just because a model was developed by a PhD in statistics and has 100 years of data feeding it does not make it necessarily predictive or reliable in today’s environment. For example, consider the insurance company that insures against natural disasters. Given global climate change that is observable over the last 20 years is the previous 80 years’ worth of data really relevant?
I would argue that risks are almost always greater than we perceive them to be and that the tail of our normal distribution is much fatter than we think. Businesses rely too much on models and should rely more on common sense. Common sense means keeping leverage low, having extra capital on hand for the next tail event, making sure that insurance policies are up to date and adequate, assessing the counterparty risk of everyone you do business with, and most importantly not blindly following a model into the abyss.