New York Times on Risk Mismanagement


Should very much enjoy a story in today’s New York Times magazine entitled Risk Mismanagement by Joe Nocera.

The story discusses risk management and introduces the concept of value at risk (VaR), an easy-to-understand measure of the risk of a portfolio of assets, pioneered by JP Morgan. The story touches on two key questions:

  • Did sophisticated risk models help avoid or rather help enable the financial meltdown?
  • To what extent should people worry about the probable 99% or the improbable 1% in assessing risk?

A few quick thoughts:

  • I found the backward-looking nature of historical standard deviation to measure the risk of a portfolio of stocks so counter-intuitive that I didn’t actually understand it in b-school until about the 3rd time it was explained to me. That is, innately, I’ve always understood that risk is about the future and standard deviation is about the past (and, in particular, the past period you are using to calculate it.) So the ideas in the story easily resonate with me.
  • The question is not whether “the math works.” The math always works. It’s about whether people understand that 1% of the time … happens, well … about 1% of the time. To me, the issue is never whether the math works, it’s about what probabilities are built into the models and what boundary conditions cause the models to become invalid. In my (semi-educated) opinion, these are always the sources of the “math problems” in finance.
  • Finally, I’ve always believed that people problems dominate the math problems. For example, in the failure of Long Term Capital Management, the root problem was that other traders started copying the arbitrage strategies they were using, effectively picking the low-hanging fruit from the risk tree. That, plus increasing hubris on the part of the firm’s principals, caused them to take bigger and bigger risks, increasingly deviating from their original strategy, and eventually leading to the collapse of the firm.


Which brings me back to David Viniar and Goldman Sachs. “VaR is a useful tool,” he said as our interview was nearing its end. “The more liquid the asset, the better the tool. The more history, the better the tool. The less of both, the worse it is. It helps you understand what you should expect to happen on a daily basis in an environment that is roughly the same. We had a trade last week in the mortgage universe where the VaR was $1 million. The same trade a week later had a VaR of $6 million. If you tell me my risk hasn’t changed — I say yes it has!” Two years ago, VaR worked for Goldman Sachs the way it once worked for Dennis Weatherstone — it gave the firm a signal that allowed it to make a judgment about risk. It wasn’t the only signal, but it helped. It wasn’t just the math that helped Goldman sidestep the early decline of mortgage-backed instruments. But it wasn’t just judgment either. It was both. The problem on Wall Street at the end of the housing bubble is that all judgment was cast aside. The math alone was never going to be enough.

The full 7500-word story is here. Enjoy!

[Addendum: a critique of the article is here on the naked capitalism blog.]

Related articles by Zemanta

2 responses to “New York Times on Risk Mismanagement

  1. Daniel Tunkelang

    I saw Taleb’s main argument as being that people forget that these models are incomplete and then substitute aleatory uncertainty for epistemic uncertainty (see the uncertainty quantification Wikipedia entry for an explanation of the difference). That’s his main criticism of the quants.But the other problem is that even intelligent people don’t do well with probabilities close to 1 or 0, even when (or perhaps especially when) those probabilities are associated with large gains or losses. That’s why people often treat highly probable events as certainties and have a hard time reasoning rationally about low-probability events like winning the lottery or being struck by lightning.And of course, everyone pays lip service to the fact that past performance is no indication of future return, but few people actually internalize it. They should remember the famous words of Niels Bohr: “Prediction is very difficult, especially about the future.”

  2. Nice comment, and thanks for the Wikipedia link to uncertainty quantification.

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.