Dissaggregation of deviations in cost analysis

The decisions on investigation of cost deviations have been generally discussed in the cost control literature assuming deviations are monitored and reported at a certain level of aggregation. This paper focuses on the relationship between the level of aggregation of the reporting system and the effectiveness of the investigating decisions. It suggests that, by incorporating into the decision process the prior knowledge managers have about the deviation sources, the deviation report could be further disaggregated leading to improved investigation decisions. This is done by using the reported deviation to update the prior distribution of its possible sources in a Bayesian manner. The resulting system is less costly to monitor than a system which monitors all the possible sources of a deviation, although it is less exact. The paper illustrates how the selection of the optimal level of aggregation should be made simultaneously with the information system and the decision rules.