CORRELATED INPUTS IN QUANTITATIVE RISK ASSESSMENT - THE EFFECTS OF DISTRIBUTIONAL SHAPE

Application of Monte Carlo simulation methods to quantitative risk ass essment are becoming increasingly popular. With this methodology, inve stigators have become concerned about correlations among input variabl es which might affect the resulting distribution of risk. We show that the choice of input distributions in these simulations likely has a l arger effect on the resultant risk distribution than does the inclusio n or exclusion of correlations. Previous investigators have studied th e effect of correlated input variables for the addition of variables w ith any underlying distribution and for the product of lognormally dis tributed variables. The effects in the main part of the distribution a re small unless the correlation and variances are large. We extend thi s work by considering addition, multiplication and division of two var iables with assumed normal, lognormal, uniform and triangular distribu tions. For all possible pairwise combinations, we find that the effect s of correlated input variables are similar to those observed for logn ormal distributions, and thus relatively small overall. The effect of using different distributions, however, can be large.