Abstract ABSTRACT: Methodologically this paper represents a synthesis and a critique of various probabilistic approaches to "return on investment" (ROI) and "residual income" (RI). Starting with assumed normality of basic underlying variables, the paper proceeds to consider more complex circumstances revolving simulation and alternatives thereto. RI is judged more versatile than ROI. Assuming normality of basic underlying variables, RI always can be assumed normal, whereas ROI cannot. This facilitates determining probability intervals through analytically derived means and variances. In more complex cases, frequency distributions available with simulation contain data for constructing probability intervals. Where simulation is not used, it is concluded that the real world applicability of Kolmogorov-Smirnov and Cramer-von Mises goodness-of-fit tests, and especially Tchebycheff-type inequalities, can be limited. Thus, high, medium and low estimates of ROI and RI are considered viable alternatives to estimating specific probability intervals.
William L. Ferrara (Fri,) studied this question.