Why there's no market turmoil on Vulcan

(Published in the South China Morning Post on 26 August 2007)

It is hard to think of something to say about the recent upheaval in world financial markets that hasn't already been said either this time round or following previous crashes.

History is prone to repeat itself, especially in the world of finance. This is because markets are the ultimate expression of human nature on a mass scale, and human nature, with its predilections for occasional irrational behaviour, remains con-

Stant.

Take humans out of markets and they would be a dull affair. Were CNBC broadcasted on the planet Vulcan, for example, correspondent Spock of Logical Securities would report: "Once again, all stocks rose today by 0.03 per cent. This is the expected rate of return and is fixed for all stocks as our perfect foresight means there is no risk. "Like yesterday, and the day before that, today's earnings results were all in line with expectations, not surprising really, given that we know exactly what the future will bring and that we all behave totally rationally."

It is our innate irrationality that drives markets. Much has been written about the role of computer algorithm traders, known as quants, in the recent selloff. Surely if computers, devoid of emotion, play more of a role in dishing out buy and sell instructions, markets should become more rational, right?

Wrong. Computer models are, of course, written by humans, and if they're all the same, which they are, herding becomes even more in-

tense.

In fact, in many ways the sharp decline in volatility and the attendant steady rise in markets in recent years were synonymous with the hypothetical Vulcan stock market. The problem was that everyone woke up one day and remembered that we lived on a planet called Earth where it was humans who approve home mortgages to borrowers with no income, humans who securitised and rated them and humans who bought them. Oh yes, and it was humans who programmed computers.

Goldman Sachs chief executive David Viniar, in a call to investors in its funds that had been damaged by their use of computer modelling; said: "We are seeing 25 standard deviation events, several days in a row." To put this into English, he was suggesting that what had happened in markets was something that should only be expected to happen every several trillion years.

Since this in itself is absurd, because it did happen; what Mr Viniar was admitting was that the models were flawed.

In this case, the computer models are not human enough. They do not take account of the impact on financial markets of their buy and sell orders, also known as the feedback effect. Self-awareness is a uniquely human trait and one that is impossible to build into computer programs given our current programming skills and computer capabilities.

There is talk of artificial intelligence and non-linear neural networks that will address this issue, but for now it remains just talk. And anyway, most of the time computer models work just fine. It's just that their very existence will mean that we'll occasionally have these increasingly-hard-to-understand ruptures in the financial markets.

It wasn't that long ago that it was the bank managers who extended loans and they knew where to find you. But times have changed and nowadays things are more complex. While this complexity on the whole improves efficiency in the financial industry, it also means that the occasional bout of panic gets amplified.

On the radio last week, a reporter asked what has become a common question — whether the rating agencies should shoulder some blame for not warning us of the imminent collapse of stock markets.

This is another absurd proposition, as it suggests that we have the capability to panic in an orderly fashion. What were the rating agencies to do? Say that everything will be fine and dandy as long as we don't rush for the exit en masse?

The fact is that the warnings were there for anyone who wanted to listen. But by then it was too late — our lemming-like mentality had taken over.

The behaviour of the ratings agencies is merely symptomatic of the problem, the root of which lies in the increased complexity of financial instruments, which in turn lies in the increased demand for risk management tools, which itself lies in our negative emotional response to financial loss, which ultimately lies in human nature.

Get the gist?

This imperfection in our psychology has been exacerbated by our desire to describe the world in a mechanistic, Newtonian way, rather than qualitatively. Thanks to Harry Markowitz et al, we have been wrongly persuaded that volatility is bad since it is synonymous with risk. His idea that investors get rewarded for assuming more risk was a sound one but his model required a quantitative measure of risk, for which he chose the standard deviation — the volatility — of historic returns.

The fact is, the uncertainty associated with market crashes cannot be defined by the bell curve as his the sudden, unpredictable lurches in volatility like we had in the past month that presents true risk not the normal sweeping ups and downs.