Reading this blog post today at the New York Times prompted me to pen some thoughts on quantitative analytics and modeling in a post-crisis era.
It is my firm opinion that adding more ‘models’ to the mix is not the appropriate response to reduce the risk of future meltdowns. The root of the issue was not that models used for valuing securities and assessing risk are inaccurate or incomplete. The issue is that the users of the models failed to ask the right questions and did not understand the answers to the questions they did ask fully as they should.
I will not sleep better at night, with greater confidence in the soundness of our financial institutions, if I am told they are all using 3 dimensional or N-dimensional models that incorporate liquidity risk and include the latest innovations from ‘adaptive-markets’ and ‘econophysics’ theory.
Increasing complexity in financial modeling has only served thus far to worsen the speed and depth of financial crises (does anyone remember the crash of ’87 and the quaint idea of portfolio insurance?). That’s not to say that more technology may not help, but at the risk of sounding Luddite, the track record thus far does inspire confidence in believing the next new wrinkle will get it right.
As Adam Smith said, the markets have an animal spirit and animal spirits are amazingly adept at confounding expectations and beliefs.
It is clear that too many of our ‘best and brightest’ on Wall Street can run computers and spreadsheets with great skill and daring. Unfortunately for those of us nursing sick 401k’s back to health, in many instances they did not understand the difference between VaR, vega, gamma, portfolio value and (insert metric of your choice here).
They did however have the false sense that they could tame the market beast because they could calculate these values on multi-billion portfolios to the penny and beyond. A laughable notion when any semblance of a market can disappear in a nanosecond.
The efforts of those contemplating the issue of ‘what do we do to make sure this doesn’t happen again’ or even better ‘how do we get on the right side’ the next time this happens (thank you Mr. Taleb for your black swans) would be more fruitful if they were focusing on the issue of improving understanding of the metrics in place now, and how can we understand and respond more quickly in times of turmoil.
I would rather see risk management and risk modelers be extremely proficient at understanding a reduced set of the appropriate metrics well, as opposed to adding layers of complexity that increase the perception of taming the beast. Let’s acknowledge that the beast is a beast and be really good at knowing how to stay on its good side.