Tuesday, March 03, 2009

Probabilistic decisions, incalculable risk

I was listening to Russ Roberts interview Amar Bhide on Econ Talk and in the course of their conversation they uncovered a few points that I found quite apropos to the current economic environment. The main topic of conversation was around Bhide's book "The venturesome economy", and many points I'll be paraphrasing.

There are two points I found particularly interesting. The first is the rise of clever statistics to create low correlation portfolios. That is, portfolio's that are no or low risk based on the model. These same clever statisticians and mathematicians created many of the exotic debt and securitized products that are at the heart of the current financial crisis. The premiss of these models is that the risk can be calculated, and thus a risk-free product or risk-free portfolio can be created. As we've discovered in since Sept 2008, when Lehman Brothers went bankrupt, these products and portfolios were far from risk-free. For me, the salient point I took from Russ and Amar's coversation, was that the heavy reliance on these statistical models, for example, the activities of the ratings agencies using mathematical models to asses risk and give AAA ratings, created an environment where due diligence was no longer necessary. At least according to the models, which convinced the market place of its low-risk claim.


The second point, is that, in reality, there are very few decisions that are actually made based on probabilities, or number crunching to arrive at data backed decision. The primary example used to illustrate this point was that of a trial by jury. The jurors are not using any probabilistic methods in making their decision. They are listening to the narrative presented by the prosecution, and then they listen to the defense attempt to poke holes in this narrative. Whether or not they find these holes in the narrative, plausible, is the fundamental question the jury must answer.

This also leads into Bhide's topic of his book, and that is Entrepreneur ship, where (some economist, I can't recall his name) defined the Entrepreneur as the individual who takes responsibility for all the incalculable risk. So similar to the jury example, an entrepreneur, has a narrative, or a belief of why their project /product/ vision will be successful, and they present this narrative to investors / venture capitalists, and if they buy the narrative, they will invest. Rarely is this decision based on probabilities or statistical analysis of the narrative.

In closing, the points that resonates with me, are the lack of due diligence in the decisions made by people at every level of this crisis. Hand in hand with this, is how easy it is to be convinced by a well constructed model that the risk has been properly hedged, yet the gap is that there are a great number of incalculable risk where probabilities may not provide sufficient guidance.

1 comment:

Jason said...

I really like this post, the narrative vs. mathematical models point resonates! I see this all the time at work, telling a story - weaving a narrative around a decision is much more powerful than just backing it with data. You often need to do both, one is not a replacement for the other. But its easy to underestimate the power of a good story!