Now, very often in the practice of economics one is overly impressed with statistics and one introduces notions like, for example, value at risk, which is the largest amount of money an entity, for example a bank, can lose within a year with a probability of 95 per cent; that is, the probability of the loss being above that is only 5 per cent. 5 per cent means this event will happen only every twentieth year, and the conclusion is that this event can be viewed as being more or less negligible. More precisely, that there is no way of getting a grasp on it. However, when I speak with people in that profession, invariably after, for example, repeating the definition of value at risk, they immediately add by saying that, "By the way, the 5 per cent of cases are the most important ones." There is a very great inequality of importance between various valuations. Ordinary valuations, which are within a few percent daily of no change, are well under control. What is bothersome, what makes the great fortunes, the great disasters, is that five percent of events. And whether it is five percent, or ten percent or two percent makes a very, very big difference. However, statistical techniques available then and to a large extent statistical techniques available today, do not allow these large values to be taken into account. To the contrary, my work in the '60s put a very great emphasis upon those values.