“…This is because the basic target of our consideration is the random sum of random variables, the distribution of which is usually far form normal. Moreover, in the previous research, Albers [1999] and Reijnen et al [2005] showed that the tail of the distribution is very sensitive to the dependence effect. It certainly is present everywhere, but in the tail it simply blows up, which can lead to huge risk underestimation.…”