How do you select a cost driver?

Identifying cost drivers can be a challenging task because there is a multitude of potential candidates. You can count the number of labor hours, machine hours or set-ups, batches processed, units produced or sold, orders, shipments, or even the number of distinct products, suppliers or customers, etc. All these measures refer to activities consuming resources and therefore may drive your costs. The more specific and homogeneous the costs you want to model, the easier the identification of cost drivers.

In addition, to guide the search for cost drivers, Activity-Based Costing organizes them in a hierarchy distinguishing between four main levels:

  • unit-level activities are performed for each individual unit; since the same amount of resources is consumed per unit, related costs are proportional to the number of units;
  • batch-level activities are performed for a group of units all at once; the same amount of resources is consumed for every batch, irrespective of the batch size (i.e. of the number of units in the batch); related costs are thus proportional to the number of batches, not units;
  • product-level activities are performed to support a specific product line, irrespective of the number of units produced or batches processed for this product line; related costs are thus proportional to the number of distinct product lines, not to the production;
  • facility-level activities are performed to benefit the organization as a whole, and as such tend to be fixed.

As discussed in Chapter 2 for the selection of a proper allocation base, two conditions must be met for a measure of activity to be a cost driver: (1) there must be an economically plausible causal relationship in which the measure of activity is the cause and the cost the consequence (theoretical causation); and (2) the measure of activity and the cost must be highly correlated (empirical correlation).

If either condition is not met, you cannot select the measure of activity as cost driver. Indeed, a potential causation can be logical but not effective in the context; therefore you need empirical evidence supporting the relation. Moreover, correlation is not causation: two variables can be correlated because they have a common cause, and not because they are related to each other; therefore, you need a compelling logical argument linking them together. Accordingly, the selection of a cost driver is guided by both a thorough understanding of operations (how are resources consumed?) and statistical tests (are cost behaviors consistent with the expected pattern?).

Solution. The number of batches processed might have had a causal relationship with machine maintenance cost, but there is no correlation here. Both the number of labor hours and the number of machine hours are highly correlated with machine maintenance costs, but the causal relationship is stronger with machine hours. Therefore, machine hours is here the best cost driver for machine maintenance costs.

Most of the time, I will assume a single cost driver. But we will also see that some of the techniques presented in this chapter can be generalized for multiple cost drivers.

Please indicate how clear and understandable this page was for you: