The purpose of multi-level contribution margin accounting in cost accounting is to analyze the profitability of products and organizational entities with appropriate allocation of fixed costs and to provide relevant information for short-term, medium- and longer-term decisions. However, the conventional framework of multi-level contribution margin accounting does not usually incorporate a mathematical optimization method that simultaneously integrates variable and fixed costs to determine the best possible product mix within hierarchically structured organizations. This may be surprising in that operations research provides an optimization model in the form of the fixed-charge problem (FCP) that takes into account not only variable costs but also fixed costs of the activities to be planned. This paper links the two approaches by expanding the FCP to a multi-level fixed-charge problem (MLFCP), which maps the hierarchical decomposition of fixed costs in accordance with multi-level contribution margin accounting. In this way, previously hidden optimization potentials can be made visible within the framework of multi-level contribution margin accounting. Applying the linkage to a case study illustrates that the original assessment of profitability gained on the sole basis of a multi-level contribution margin calculation might turn out to be inappropriate or even inverted as soon as mathematical optimization is utilized: products, divisions, and other reference objects for fixed cost allocation, which at first glance seem to be profitable (or unprofitable) might be revealed as actually unprofitable (or profitable), when the multi-level contribution margin calculation is linked to the MLFCP. Furthermore, the proposed concept facilitates assessment of the costs of an increasing variant diversity, which also demonstrates that common rules on how to interpret a multi-level contribution margin calculation may have to be revised in some cases from the viewpoint of optimization. Finally, the impact of changes in the fixed cost structure and other parameters is tested via sensitivity analyses and stochastic optimization.

No abstract

No abstract