Broken models and the new world order

The last two days of
the week just gone, I was in Barcelona at an event called Smart
Banking Forum, run by IBM. The speaker that stood out most in my mind
was Shankar Ramamurthy, a senior IBM thinker in their Financial
Services consulting practice.

Shankar was commenting
on the current state of financial markets, and made a number of
(probably) contentious statements, which nonetheless got me thinking.

His first was that it
is reasonable to expect an increasing frequency of globally
significant market events (such as the present crisis we are
experiencing). He describes these as “six sigma” events, and
suggests that our existing models for (amongst other things)
catalouging and pricing risk, are out of kilter with this new

He went on to suggest
that the Efficient Market Hypothesis is now bankrupt and that the
underlying assumptions of a normal distribution in all our risk
models are breaking down in the face of all this.

Shankar went on to say
that markets are increasingly described by a “fat tail” rather
than a standard normal distribution, and that the maths involved in
this description are presently not well understood by banks, or
anyone else for that matter.

Now this, if it is all
true, is quite bad news.

Let us put aside the suggestion that everyone is tied to normal distributions for a moment and concentrate on some of these other points.

If we assume the basic
maths that underpin much of what we do is now invalid (or at least,
or at least an incomplete description of what might happen),
there is a huge threat that in this new world, we'll keep getting hit
by crisis after crisis, and the whole system will collapse. In
Shankar's typology, there will be an increasing frequency of “six
sigma” events, and noone will be able to see them coming in time to
do anything about them.

of course, also presents a significant opportunity. Banks with back
rooms filled with mathematical experts have a very new, and very big
challenge on their hands. Those who crack the maths first will be
able to play at the very edge and make windfall gains. But, lets face it, if Shankar is correct,  the
fundamental research to exploit the opportunity has presently not been done.

the question in my mind is whether the scenario being painted by
Shankar is, in actuality, an artefact of too narrow a focus on
specific markets. As he says, markets have memories, and furthermore,
are interconnected to an increasing degree as a result of
globalisation. Might we, therefore, expect that models with historically limited scope might fail to capture new, emergent effects resulting from the new world order?

existing models, rather than abandoning them altogether seems a more
likely prospect for most institutions, as far as I am concerned. That
seems a project that single bank could undertake without tearing
itself apart.

of this is predicated, of course, on the view that something
actually broken at present. And to this last point, I am somewhat
unconvinced. There have banking failures before, and there have been
significant (and unexplainable by Efficient Market Hypothesis) events

is going on here which is so astoundingly new in the overall tapestry
of the history of financial services? Granted the specifics of the
current situation might be novel, but is this crash really any
different to the ones that precede it?

My suspicion is that there is no new world order after all, and that things will move forward in the ordinary way again relatively soon.

4 Responses to“Broken models and the new world order”

  1. Craig
    October 13, 2008 at 9:49 am #

    The very troubling thing is that if this do go back to ticking along nicely we won’t know if 1. all is ok and it’s just another crash or 2. everythings broken and there’s a hidden precipice just ahead.
    The clear message must be that the research must be done to review the current models and the current data.

  2. October 13, 2008 at 3:15 pm #

    I quite agree. The research we have today isn’t necessarily conclusive. But who is going to do this research, and how long will it take? It took decades to build and perfect the models we have today. If, in fact, the world is really ending, there aren’t decades to build the models available.
    Then it really will be back to basics banking, which means, in the end, less access to banking for most people.

  3. October 13, 2008 at 3:22 pm #

    The models have been broken for quite a long time, people have written books about it. notice the price !
    (I don’t think normal distributions are being used by anyone these days though.)
    The “six sigma” events are going to be natural disasters, political collapses and wars. Maybe banks should hire more historians!
    It would be in everyone’s interests for things to settle down, after our little global “Minsky moment”, but I read The Great Illusion at uni, so I’m not so sure it will happen…

  4. David Shea
    October 14, 2008 at 3:05 am #

    When do you get burned? When you don’t realize something is hot or you forget. Everyone learns from experience and this has been a learning experience. Add it to the model and try not to forget how much it hurt. 😀

Leave a Reply

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


Proudly powered by WordPress   Premium Style Theme by