THE STRAW THAT BREAKS THE CAMEL'S BACK
By: Robert Wagner
April 10, 2017
In my last note I explored the nature of certain types of risks, which tend to gradually build up and avoid detection until it is too late. As I thought about this type of risk further, I wanted to revisit some aspects of the "frog problem". As I thought about how to frame up these ideas, I decided that perhaps one way to elaborate on some key points is to dedicate this note to 'the camel'.
We have all heard many times the idiom referring to the straw that broke the camel's back. This is precisely the point where the frog finally gets cooked and the camel collapses under his heavy load. The definition generally describes the cumulative effect of small changes which eventually become overbearing. Any one piece of straw by itself is meaningless, almost laughable compared to the strength of the camel. However, when many thousands of pieces of straw get bunched together the pile takes on properties distinctly different from its individual components. The final straw is the very last minor change which creates a dramatic reaction.
I witnessed an example of such a process a few years ago in my own front yard. Our neighborhood is relatively older with many mature trees covering the area. A major branch from one of the large trees in our yard suddenly crashed down on a clear, calm, and sunny summer day. There were no warning signs. The tree was large, fully covered with leaves, and has not lost any limbs during the periodic thunder storms that are typical to the area. We were lucky, and no one got hurt. But it got me thinking what led to this sudden dramatic development on a tree that probably stood there for over eighty years. Could it be that one small bird landed on the wrong branch, or one of the squirrels added some extra weight to unbalance the tree? There was not the slightest wind, nor any other visible disturbance to provide an obvious explanation.
Once the dust settled, it became apparent that this old tree, seemingly healthy, was in fact rotten to the core. Every day of every year it continued to rot just a little more until it became so fragile that perhaps an otherwise minuscule factor caused it to tip over and crash to the ground. We subsequently called in an arborist who recommended taking down several other large trees which otherwise looked fine on the outside.
How could one get smarter about recognizing these types of risks? Lets revisit the camel as an example. We could start by learning some basic facts about this type of animal. A typical 'hybrid' camel weighs about 650 kilograms, and has a carrying capacity of between 400-450 kilograms. You put a small piece of straw on his back and it will not even notice. You begin to add some more, and at some point approach the 400-450 kilogram range. At that point the camel definitely notices the load, but can comfortably carry it day after day.
If you were to keep loading more and more straw on the camel it would start to get very uncomfortable. I don't know what the breaking point could be but I am fairly certain that it could not carry a full metric ton. Where then is that last straw, and when do we pass from the safe zone to dangerous territory?
At first look it seems to be very difficult to predict which piece of straw will be the last one. One could model the problem by including several key variables such as: how old is the camel and what shape is it in; is it standing still, walking, or running with its load; what are the conditions in terms of temperature, wind, humidity, etc. We could build a complex Monte Carlo simulation model and begin to estimate the odds of each piece of straw being the one.
Alternatively we could start with our knowledge or experience about the capacity of a typical camel. Up to 400-450 kilograms we should be reasonable sure it can handle the load. At that point we should forget about individual pieces of straw. Think in bigger chunks. Can we add another 100 kilos? Perhaps the right camel under the right circumstances may be able to handle that just fine. Can we add 200 kilos? Now the camel is really stressed. Maybe there are some who could still do it, for short periods of time. Can we add 500 kilos? The answer seems to be most likely not.
Maybe the point is that it is somewhat futile to even try to estimate which piece will be the final straw. It will be inherently random. And perhaps we shouldn't even care. What seems to matter is to develop a sense of an acceptable range, a grey area where we may be able to increase load with careful consideration, and a clear limit after which the risks become too great and we simply have to stop. Instead of building more and more precise models to offer incrementally better and better estimates of risk, we create some simple guidelines to keep us out of trouble.
Problems involving risks that share characteristics of our camel example can be solved with some study of the "fundamentals" and developing relatively simple guidelines or ranges to work with. As the pressure builds, thinking in more sizable chunks rather than small pieces of straw can also help raise awareness of the gradually building risk by consciously making it less gradual. There are many practical examples of this type of risk in business, government, and various other types of settings.
One consideration that we have to accept is that under these guidelines the camels will likely not be fully optimized. They will carry less than they perhaps could. But they will all live and continue their work rather than eventually collapse due to overload. There could be criticism that the guidelines come at an opportunity cost, and these arguments may need to be examined case by case. Unfortunately some camels' backs may have to be broken before real attention is paid.
Now, if we depart a bit from our camel, zoom out, and think about the broader context of slowly creeping and accumulating risks, the picture gets a bit muddier. The camel problem is easy in a sense that we already know about the straw, and the camel, and know what we don't know (which piece is the last straw?). It is a question of quantifying a known risk, and deciding the best approach for that. We should look around and see where we can identify 'camel-like' circumstances where developing guidelines could be a reasonable approach.
What about many other potential risks that are not so clear... the famous "unknown unknowns"? Think about my tree that tipped over. One second before it happened, I was ignorant of the risk, and I was doing yard work not far from that tree. The risk was there slowly building up over time, but it remained invisible to me. So how does one look for the proverbial rotten tree before it falls on top of us? I believe part of the answer lies in carefully evaluating not only our estimates of risk, but also paying close attention to the size of underlying exposures, and really understanding the difference between the two. The topic of risk vs. exposure will be the subject of my next note.
