Predicting random shocks

DESPITE nearly 100 years of development in econometrics, the treatment of random shocks remains quite naïve. Economists overuse normal distribution, the classic bell curve, to model the influence of variables, filtering out shocks. The logic is that most shocks are small. Larger shocks are less likely. Standard deviations measure departures from the average. Almost all shocks, 95 per cent of them, are within two standard deviations of the average.

But that model may be deceptive. Benoit Mandelbrot, the French mathematician, tried for decades to convince econometricians that the distribution of random variables, shocks, may be non-normal. He was ignored as either too naïve or too sophisticated to be useful. The efficient markets hypothesis, promoted by the economist Eugene Fama, prevailed.

The 2008 financial crisis blew a big hole in the efficient markets hypothesis. A massive shock shook the stock market. If Dr Mandelbrot's non-normal distribution model was followed, it may have been anticipated.