How much is that doggy in the window? A reflection on the Crash of 2007

By | February 16, 2012

Chris-Bentley_MBA-2006Chris Bentley graduated from his MBA at Bradford University School of Management in 2006 and joined Holidaybreak plc, where he is now Head of Corporate Development. He has supported the company’s strategic development, leading the Group’s merger and acquisition activities. He also chairs the company’s Adventure Travel Division, and works with business units on the evolution and redesign of their business models. Here, he shares his views on a recent guest lecture by fellow alumnus Terry Carroll.

I’ve started to make a point of attending the Guest Lecture series at Bradford University School of Management whenever my schedule permits.  The latest, Terry Caroll’s ‘2007: How did it happen?’ was a brilliant and relentless exposition of the myriad factors that combined to make the ‘credit crunch’ such a perfect financial storm: the housing market factors, the financial market factors, the regulatory factors, and so on.  There was one issue though which Terry hinted at and then veered away from.  Understandably, as we’ll get to in a minute.  But, to me at least, it’s a point which may be at the heart not just of why there was a crash, but of why it came so close to annihilating the global financial system altogether.  It offers hope, but it also points to some unpalatable questions about the future of capital markets.

At the end of Terry’s lecture, Dr Peter Prowse made the point that we seemed not to have learned from a long history of bubbles and crashes.  And he was right.  Bubbles will always be with us as long as there are feedback loops.  Heck, all it really takes to stoke a property bubble is a plentiful mortgage market and a bit of exuberance.  But bubbles have burst before without causing anything like this amount of mayhem.  As one audience member asked, what was it that was different this time?

By way of answer, Terry pointed to the massive growth in derivatives, dubbed ‘financial weapons of mass destruction’ by Warren Buffet.  True, I think, but specifically why?

Another audience member asked whether there were any ‘red flags’ that could have warned us in advance.  Terry pointed to several, including the now famous economist Nouriel Robini, ‘Dr Doom’, who was predicting a housing market crash and global recession as early as 2005.

There was another prophet though.  And his red flag was in the air almost half a century ago.

How much is that doggy in the window…

The gasoline that fuelled the crisis combined global financial imbalances, lax monetary policy, the bubble in the US sub-prime property/mortgage markets and, what was it…?  Oh yes, of course, ‘greed’.   But the spark that ignited it was buried in the derivative securities into which mortgages were bundled and sliced into tranches.  Each tranche was designed to exhibit a specific profile of risk.  Putting a price on a tranche was basically about pricing that risk.

…the one with the waggly tail?

Which is tricky.  What is the most likely level of loss within a given tranche, and what is the ‘tail risk’ that losses may be significantly higher than this?

It turns out that everyone was pricing risk in the same way, using a statistical model with origins in the actuarial sector, called a Gauss copula.  The model attempts to take account of simple correlations between variables, but at heart it is still a ‘normal’ distribution, a bell curve, with events clustered fairly tightly around an average and tail risk falling away sharply either side of the mean.

In 1962, a deeply eccentric mathematician called Benoit Mandelbrot showed that fluctuations in the cotton market were anything but normally distributed.  His ideas caused some excitement through the 1960’s, but had fallen out of fashion by the seventies.  In 2004 he published an exposition of his earlier and more recent work and a summary of research by others, including Fama, Schoutens and De Vries, all of which pointed to massively higher tail risk than would be implied by Gaussian models, in markets as diverse as blue-chip US stocks, the S&P 500, the sterling-gilder exchange rate from 1609 to 2000, 19th Century railroad shares, US Treasury bills, call money (interest rates on loans from banks to brokers), gold prices and foreign exchange.  To put these waggly tails into context: one study by Citigroup in 2002 found gyrations in dollar-yen rates as high as almost eleven ‘standard deviations’ – something that should not have happened even once  if they’d traded dollars and yen every day since the Big Bang[1].

The Gauss copula models being used by everyone, including the credit rating agencies, were massively underestimating the degree of systemic risk in the sub-prime property market.  In one of life’s great ironies, the Gauss copula model itself, and especially its universal adoption, was probably no small contributor to the systemic risk that invalidated it.  The sub-prime dogs would turn out to have very fat tails indeed.

As the sub-prime market started to come off, even the ‘safe’, AAA-rated tranches began to take losses, a near impossibility according to the financial models.  And because everyone was using the same model, which clearly didn’t work, no one had a clue anymore how much their portfolios were worth or, more to the point, how much trouble other people’s portfolios might be in.  And in conditions of such near blindness, banks simply stopped lending to each other.  The vital interbank funding market, the beating heart of the global financial system, ceased to function overnight.  And civilisation found itself staring into the abyss.

I suppose that it’s too much to ask of our politicians that they try to understand and explain this stuff.  It’s hard to imagine even David Cameron’s oratorical skills stretching to a passionate advocacy of the Lévy stable distribution as the solution to our ills.  But the alternative is that banking reform is sold to us without a complete understanding of what went wrong in the first place.  Worse, as an expedient substitute for a proper explanation, an entire class of people (not just culpable individuals) has been vilified.

Why didn’t the rocket scientists in the middle of all this realise what was going on?  There were competing theories, and there were some exceptions to Mandelbrot’s findings.  The financial economics of asset and derivative pricing wasn’t (and still isn’t) black and white.  And then there is the funny thing about risk – being prudent can be very costly.   If you’d taken the Mandelbrot view in 2002, while everyone else was going gangbusters, your shareholders would have fired you for incompetence by 2007, as your investment returns trailed in the dust.

Benoit Mandelbrot died on the 14th of October 2010, aged 85.  His views offer hope – it is hard to see that puppies will be quite so mispriced in future.  But they also raise big questions – not least that of how capital should be allocated in a world where tail risk dominates.

[1] The (Mis)Behaviour of Markets: a Fractal View of Risk, Ruin and Reward, Benoit B. Mandelbrot (2004)

Click here to view a video of Terry Carroll’s recent guest lecture on ‘2007: How did it happen?

Click here to read Terry Carroll’s blog on careers advice for ambitious undergraduates

5 thoughts on “How much is that doggy in the window? A reflection on the Crash of 2007

  1. Friedrich-Joachim Sack

    Excellent post, which has indeed provided me with some completely new aspects. I personally believe that situations will always be -in the majority of cases at least- so complex and unforeseeable that they cannot be measured by one single model but only by combining the outcomes of several models simultaneously. And even that may still be characterized by major uncertainties.

  2. Steven Olivant

    Chris gives a clear explanation – and a useful reminder – of why tail risk is more dangerous than many people think.

  3. Adrian Nixon

    A fascinating post Chris,

    – Everyone making decisions based on the same information.
    – Analysed with the same (flawed) tools.
    – A business culture that rewarded risk taking and punished prudence
    – Led to chaos

    Leaving us with a thought provoking legacy of how to price tail risk.

    As you clearly point out, the rush to banking reform is being built on very shaky foundations.

    There are deeper changes in the pipleline for the world economy.

    I fear this may not be the last crash.


  4. David Bagley

    An interesting article, Chris; I wish I had been at the lecture and could comment in a more informed manner. It does occur to me, however, that for a given set of circumstances, “the market norm”, financial transactions will be tightly focused around the mean in a normal distribution as you describe. Unfortunately, during the credit crunch, as fear began to grow in the market around the disposition of holdings in sub-prime assets and the many derivatives and sub-derivatives based thereon, the market norm began to break down and panic took over. Pricing of financial assets moved decisively to the left as people sought to dump assets almost at any price driven both by concerns over counterparty risk and their own inability to meet day-to-day liquidity requirements given the inability of the interbank market to provide the short term liquidity required by the system. This drove extreme selling behaviour. Too many sellers and not enough buyers and the prices can end up anywhere – and they did. But, when the dust settles and some equilibrium returns, I suspect that the standard deviation model will still apply albeit around a new lower mean and what we saw in the darkest days of the credit crunch was a lack of equilibrium between buyers and sellers due to inadequate information about a range of issues but primarily counterparty risk.

  5. Chris Bentley

    Many thanks for all your comments.

    @Friedrich: Multiple models at multiple levels I guess, which is presumably the logical outcome of adopting a ‘macro-prudential’ approach to regulation. In business strategy we’re quite used to analysing macro, micro and business model structures and figuring out the linkages between them. Taking account of feedback effects is a natural thing to do. Not so with the MBS sector apparently.

    @Adrian: Indeed. A big problem is figuring out secondary and tertiary, ‘butterfly’ effects. The fact that a country representing, what, 2% of Eurozone GDP is causing global economic disruption and may yet collapse the Euro is arguably a traceable consequence of the sub-prime crash. I know that you have some head-spinning observations from your own work that may implicate certain oil supply fluctuations in the same way…

    @David: Don’t forget that Mandelbrot’s research long preceded the credit crunch. If financial market outliers are both more probable and of higher impact than Normal models would predict (the suggestion being that a flatter, ‘power law’ model is more relevant), in both cases because of feedback loops and complex systemic effects that invalidate Normal assumptions, then the only logical conclusion is that any investment should face very much higher hurdle rates than we’ve been used to, with potentially catastrophic effects on industrial investment. And probably that we should put our personal savings into agricultural land. So I dearly hope that you’re right.

Leave a Reply

Your email address will not be published. Required fields are marked *