Risk and Uncertainty

Not too long ago, I had an argument (discussion?) with two other people, on an issue regarding risk management. I maintained that despite all the actuarial science that insurance companies utilize, they still tend to underestimate risks due to the uncertainty factor. By uncertainty, I refer to events that are unpredictable, events that are not covered by past history, and hence left out in the calculations so integral to the decision-making process. As a result, with the underestimation of risks, insurance premiums are lower than they should be, and companies can go bust quite easily. I cited the example of 9/11 to illustrate my point, where several insurance companies posted enormous losses because they did not factor in the possibility of 9/11 occurring.


Meanwhile, my two contemporaries dispute my point, claiming that insurance companies do know the risks involved, that their actuarial science is impeccable, and their models solid. The fact that so many insurance companies went bust because of 9/11 is not because of a flaw in their calculations or because they failed to factor in uncertainty, but because they are “suay”.


Apparently, according to them, if your model states that there is a one-in-a-trillion chance of this event occurring and this particular event still happens, you are just “suay”. The model is correct. The chance is still one-in-a-trillion. The fact that it happens simply means that you are unlucky.


I disagree. For me, when it is a one-in-a-trillion chance (statistically insignificant) and the event still happens, obviously something is wrong with the model and it needs more work. Risk premiums need to be adjusted upwards. More factors need to be considered.


In the end, none of us succeeded in convincing the other party.


To further substantiate my argument, I would like to borrow an example from The Black Swan by Nassim Nicholas Taleb. Imagine yourself as a turkey on a farm, well-fed and well-cared for all throughout your life. Nothing negative has ever happened to you, and using past history, you extrapolate and think that nothing bad is ever going to happen. Meanwhile, Thanksgiving Day draws nearer and nearer. What is the probability that you are going to get hauled off to the slaughterhouse in the next few days? By using past history, the probability is zero. Nothing bad has ever happened, so why should it happen now? But in actual reality, with each day that passes, the probability jumps exponentially, until eventually, Thanksgiving Day arrives and the unsuspecting turkey is in for a nasty surprise.


Risks are quantifiable. Uncertainty is not. Life is about uncertainty. Please don’t torture reality to fit your “models”.



Or maybe, I’m just a cynical skeptic who doubts everyone and everything.