You are here:

Should risk managers be sacked?

Created:
18 September 2008
Written by:
Chris Dillow

In an interview here (MP3), Nassim Nicholas Taleb, author of The Black Swan, says financial institutions have shown down the years a "proven total inability to make money on risk-taking", and so should fire all their risk managers. Is he right?

Advertising

I've some sympathy for this view. I've always thought it odd for risk management to be a separate function, simply because risk and return are two sides of the same coin; if you want to find a high-return strategy, you should look for a high-risk one. Separating risk management from trading is therefore ensuring that the left hand doesn't know what the right is doing. It's a recipe for introducing yet more agency problems into investment banks - as if they didn't have enough already.

And I think Taleb is right on other points: to warn that history gives us no idea about tail risk; that the Gaussian assumption that returns are normally distributed is "idiotic"; that single measures can't capture uncertainty; and that attempts to do so can lead to a false sense of confidence and hence to excessive risk-taking.

But, but, but. I suspect - and certainly hope - that Taleb is attacking a straw man here. Risk managers must, or should, know all this. We all know that extreme returns are not normally distributed; I learned this on Monday 19 October 1987. Indeed, research in recent years suggests such returns follow a robust statistical law - the cubic power law, discussed by Xavier Gabaix here. We know too that apparently uncorrelated assets have a nasty habit of becoming correlated in bad times, either because of contagion or because a drying up of liquidity in one market forces investors to try and sell other assets. We learned this when Long-Term Capital Management collapsed in 1998.

Anyone with an ounce of wit should have worked out that credit derivatives wouldn't be immune from these difficulties, and that it was dangerous to price them on the assumption they would be. As long ago as 2001, this paper showed that: "The assumption of a Gaussian copula may not adequately model the potential extreme risk in the portfolio."

And we have known for decades that there's a distinction between risk ("known unknowns") and uncertainty ("unknown unknowns"). Frank Knight made the distinction back in 1921 in his thesis Risk, Uncertainty and Profit . Anyone with a passing acquaintance with Keynesian economics will know this.

So, what Taleb is attacking is not risk management as such, but bad risk managers. And let's all hope there's a huge difference - because the thought that risk management is being done by people even stupider than me is too terrifying to contemplate.

MORE FROM CHRIS DILLOW...

Read more of my musings at www.investorschronicle.co.uk/chrisdillow.

A selection of my favourite blogs, and data sources, appears under 'External links' on the right-hand side of the page.

I moonlight in the blogosphere, too: http://stumblingandmumbling.typepad.com

  • Order reprints
  • Back to top

Login

Login

Forgotten password?

FREE ANALYSIS EMAIL
  • Get our FREE daily investment email. Informed comment on strategy, shares, funds and derivatives. Direct to your inbox at 3pm every day.
Free daily e-mail