Rolling the dice on the future of the economic crisis

Consider a couple of dice games, brought to my attention by the Reuters economist John Kemp. In the first dice game, the player wins if he rolls one to five, but losses if he rolls a six. For any individual roll, the probability of winning is about 83 per cent, and the risk of losing is about 17 per cent. So long as the player can roll the dice repeatedly, over the long run these odds will be proven correct.

The financial world works on this basis. Banks and the like effectively predict the likely wins and losses from future die rolls by observing the outcome of thousands of past rolls and assuming the game parameters will remain the same.

But imagine a second game in which the player still loses on rolling a six, but rolls using a die with an unknown number of sides.

Suddenly, the outcome of an individual dice roll cannot be predicted in advance. The whole structure and parameters of the game are unknown.

Without knowing the parameters and the probable distribution of outcomes between wins and losses, it is impossible to model long run returns statistically. The odds are simply unknown.

As Kemp observes, the first game is characterised by risk, while the second game is characterised by uncertainty.

"Risk is fairly easy for financial institutions to cope with. It can be modelled by probability distributions, priced and transferred. Insurance companies and banks have been doing this for hundreds of years; the same techniques are the foundation of the modern securitisation industry.

"Uncertainty is much more difficult to handle. When parameters are unknown, investors cannot calculate likely loss rates or how much yield/premium to charge in order to cover losses in the long run."

You will guess where we are headed here. The key problem that we have been facing since the onset of the credit crisis is a huge increase in uncertainty about the future rather than simply an increase in the riskiness of financial markets. Kemp suggests we are working with a "multihedral" die. It doesn't sound much fun.