For purposes of exposition and application, a linear programming formulation of the planning process was postulated. Recent application of the proposed approach indicates that this is a feasible approach to generation of control information. And, in fact, the application's results are highly encouraging. First, important deficiencies in the traditional accounting system were observed. Since this traditional model is not an opportunity cost system per se, the most it can be expected to do is to signal the existence of opportunities. In the period analyzed it did signal the existence of deviations although their source and effect were obscured. A second, and more subjective result, is that application has given every indication that this type of approach is indeed feasible (which should not be confused with an assertion that it is desirable from a cost and value viewpoint). Consequently, it is appropriate to suggest that additional research be done in an effort to examine such issues as the extent to which accounting variances should be aggregated, given a specific decision model, or the extent to which decisions made in response to certain deviations should be considered in the accounting system.

This article focuses on a study which demonstrated how similar types of standard cost variance analysis may be applied to the control of the application and performance of formal decision models. Previous efforts to improve the usefulness of standard cost variance analysis have tended to focus on one of two related problems. Proposals have been offered that had as their objective either an improvement in the types of variances the accountant should calculate or in his methods for analyzing the significance of observed variances. The essential difference between process control and model control lies in the type of response management should make to an observed variance. The appropriate response by management depends upon the expected source of the deviation. Traditionally, standard cost systems have concentrated on the analysis of type 1 and type 2-a deviations. These systems have not been designed to indicate when and if type 2-b deviations are critical to the decision process.