Intelligent cost system design.
Prior to the advent of modern computers, the collection and manipulation of data was extremely expensive. To avoid excessive measurement costs, cost systems were designed using a minimalist approach. This trade-off led to poorly designed cost management systems that relied on data that were already available and therefore did not have to be specially collected and manipulated. With the dominance of the scientific management movement, the time it required the direct labor force to complete tasks was tightly monitored, often in minute detail. As most manufacturing processes were labor intensive at the turn of the century it was only natural to develop labor-based cost systems that assigned all indirect resource-costs to products (or, more rarely, services) using direct labor hours or dollars. The only other cost assignment approaches that were readily available were number of units and material dollars. Number of units is appropriate only when the products are essentially identical, so it's rarely used outside process costing. Material dollars is typically a poor predictor of the indirect costs of the products, so it's rarely used other than to assign procurement costs to products in the most sophisticated of systems. Thus, direct labor was the best choice of the limited set of cost drivers available and hence dominated early system design philosophy.
The result of this trade-off was a preponderance of cost systems that relied on direct labor to assign indirect costs to products and services and used a very limited number of cost pools to achieve that objective. Only in the simplest of manufacturing settings will such systems be able to provide adequate insights to managers about the costs of their products and how to reduce those costs.(1) These systems can be viewed as being designed under minimalist principles, which assume high cost of measurements. Many accountants still assume that the cost of measurement is high and subconsciously adopt a minimalist design philosophy. It is these accountants that in the early days rejected (and in some cases continue to reject) - without appropriate analysis - activity-based costing as being too complex and costly.
As information technology reduced the cost of data collection and manipulation, the ability to design more complex, cost effective cost systems led to the emergence of multiple cost pool designs. Instead of facility-wide or major area-wide (such as automatic machining and electroplating) cost pools, firms began to use machine-class cost pools to provide them with more accurate product costs and, in particular, enhanced cost control.(2) The emergence of highly automated machines, which replaced direct labor with support labor, reduced the accuracy of direct labor as an estimator of indirect costs. This reduction first led to the emergence of machine hour-based cost systems. But these cost systems, while more sophisticated and accurate, still relied on unit-level cost drivers and were still traditional cost systems with all of their limitations in handling the costs of batch- and product-level activities.
It wasn't until the 1980s that systems that used non-unit level cost drivers began to emerge. These systems reflected the interplay of four interrelated changes in production settings. First, the level of competition that most firms faced was increasing due to the emergence of a global economy. Typically, as the level of competition increases, profit levels fall, and it becomes more important for the firm's cost system to report accurate product costs. More accurate products costs are required for the firm to fine tune its product mix so it's inherently more profitable. Second, to remain competitive many firms were increasing the complexity of their product mixes. The adage, "You can have it in any color you want as long as it's black" changed to "You can have it in any color you want." When product complexity increases, so typically does the level of batch- and product-level activities performed. Third, the increase in the level of automation also increased the relative (and probably absolute) level of batch- and product-level activities as the direct costs of products fell. Finally, the spread of information technology meant that the range of potential low cost to use cost drivers was expanding rapidly. In the early 1980s, these trends both required and enabled cost effective activity-based cost systems to evolve. These systems appeared simultaneously in America and Europe.(3)
But as system designers began to shed their minimalist assumptions, many went into overkill and adopted a maximalist perspective. They began to design overly complicated activity-based cost systems that were supposedly able to provide accurate product costs and support operational improvement and learning. The problem with the maximalist approach is that there is almost no limit to the number of activities that can be identified. For example, a setup can be seen as a single activity, "setup," or a series of activities such as "break-down," "tool insertion," "setup," and "first inspection." Furthermore, each of these activities can be associated with a unique activity cost driver, requiring considerably more measurements be undertaken. The result is excessively complex activity-based cost systems that are indeed too expensive to maintain.
The problem lies in the confusion between the appropriate design of systems whose primary objective is to support strategic product costing and those whose objective is to support operational improvement and learning. For product costing purposes, relatively simple activity-based cost systems will suffice. For example, in most settings only a few unit-level, batch-level, and product-level activities need to be identified. The reason simple systems will suffice is that strategic costing typically requires only an accuracy of [+ or -]10%.
Consequently, consolidating multiple activities into a single cost pool and using a single activity cost driver to assign costs to products is acceptable. Furthermore, the costs of the resources consumed by these activities can be assigned periodically, thus reducing the number of times per year that data have to be collected and the system run. In addition, indirect cost assignment is acceptable for strategic costing. For example, the total machine electric bill for the facility can be assigned using the product of machine hours and machine horsepower. The resulting product costs will be estimates of actual electricity consumption, which is more than adequate for strategic costing purposes.
For operational improvement and learning, the number of tasks (a term we will use for more detailed activities) to be identified is large. The difference stems from the objective of the system. Operational improvement requires that the feedback provided enable the individual to change his/her behavior in ways that lead to increased efficiency. For the feedback to help achieve that objective it must be quite detailed. Thus, more tasks than activities have to be identified in the strategic costing system. For example, for the purchasing department, the only strategic activity might be "cutting purchase orders." For operational improvement and learning purposes, the tasks might include not only "cutting purchase orders" but also "certifying vendors" "writing contacts," and "identifying new vendors."
Furthermore, while a strategic costing system can rely heavily on indirect cost assignment, for operational improvements such estimates are worthless. The individuals whose performance is being evaluated will simply argue that the estimates are incorrect and insufficient for performance evaluation, and they'll be right! Performance evaluation requires real-time direct cost assignment. For example, to adequately measure electricity consumption improvement, you must measure the actual consumption of each machine by product. Providing this information as feedback enables each individual to be held responsible for his/her actual resource consumption.
These two differences - more detailed task analysis and real-time direct cost assignment - make operational improvement systems more expensive to design and maintain than strategic costing systems. When activity-based systems are designed that purport to achieve both objectives, maximalist systems usually evolve. The intelligent design philosophy is to take a two-system approach. The first system is a strategic cost system that is as simple as possible and based upon activities that number in the tens not hundreds. This system is used to report product costs and hence support strategic decisions. The second system is designed to support operational improvement and learning. It is also designed under an "as simple as possible" philosophy. But as this system provides the greater detail and direct assignments required for continuous improvement and learning, here the number of tasks can easily measure in the hundreds. The key to intelligent design is the scope of the two systems. While the strategic system would encompass the entire firm, the operational improvement and learning system would encompass only those areas of the firm where activity-based costing support for operational improvement and learning is effective. It is this partial coverage that enables the operational and learning system to be as simple as possible and not a maximalist design. Thus, the operational improvement and learning systems wouldn't be a "complete" system from a firm-wide perspective but would reflect intelligent design. The design of such double systems is on the frontier of modern design theory. That will be the focus of the next column.
1 This statement isn't strictly true. Under the scientific management movement, the aim was to achieve consistent performance, not to continuously improve performance. Prior to the emergence of continuous improvement, the main aim was to control costs. Today the main objective is to reduce costs.
2 German cost systems have used machine class cost pools since the 1920s. They typically are very detailed and used to control costs to a high degree.
3 There is no reason to suspect that activity-based cost systems didn't appear before the 1980s. But there's no documentation of extensive application of the principles before that time.
Robin Cooper is professor of cost management at the Goizueta Business School at Emory University.
Regine Slagmulder is professor of management accounting, Tilburg University (the Netherlands), and visiting professor at the University of Ghent (Belgium).
|Printer friendly Cite/link Email Feedback|
|Author:||Cooper, Robin; Slagmulder, Regine|
|Date:||Jun 1, 1999|
|Previous Article:||Process improvement = productivity gains.|
|Next Article:||Qualified tuition savings.|