A General Approach To Modelling Non-normally Distributed Cost Variance Data: An Illustration

Abstract

Many studies have indicated that many financial and cost variables are non-normally distributed. For example, empirical studies of financial ratios and cost variance data have found non-normality. However, instead of identifying the correct distribution of these financial variables, many of these decision models (e.g. bankruptcy prediction, bond rating prediction, cost-volume-profit analysis and the cost variance investigation decision) have simply adopted an assumption of normality. This is contrary to the empirical evidence which has suggested that some of these input variables are non-normally distributed. The reason for this common adoption of the normal distribution assumption could be attributed to the lack of knowledge regarding the appropriate distribution function for the variable. In order to encourage the use of the appropriate distribution function for the required financial variable, this paper examines this aspect. The objective of this study is to assist in identifying an appropriate distribution function for modelling non-normally distributed accounting variables. A three-step procedure is described and actual cost variance data of a Fortune 500 company's manufacturing plant are used to illustrate the procedure.