Evaluating and choosing cost drivers Cost functions have a good fit either simply by chance or because there is an underlying economic relation between the cost and the driver. The cost function is more likely to predict accurately in the future if that function is based on an economic relation. It is important for cost functions to be economically plausible, in addition to fitting the data well. Some students give excessive (sole) emphasis to the goodness-of-fit criterion. For example, when using simple regression (where one independent variable is used to explain the change in the dependent variable, students often choose the variable that has the highest r 2 . Stress the need to consider all three evaluation criteria. As inmost real-world applications, the cost function is not valid at shutdown in the Møre-Teppe example. The intercept term does not capture fixed costs at zero machine-hours, because at shutdown, some costs can be avoided e.g. companies can generally cut FC by dismissing salaried workers, selling PPE, etc. In sum, zero volume or shutdown is typically outside the relevant range, so the intercept term should not be interpreted as fixed costs at shutdown. It is simply the constant component of the equation that provides the best (linear) fit. Students are often confused by the difference between the behaviour of total costs and that of unit costs. Exhibit 9.8 illustrates a cost function that increases at a decreasing rate. Stress that a declining cost per unit does not mean that total costs decline. (For total costs to decline, unit costs would have to become negative) With a declining cost per unit, total sales will still increase, but at a decreasing rate as in Exhibit 9.8. To illustrate, ask students to think about purchasing personalised business cards. Cards might cost €50 for the first 100, €40 for the next 100 and thereafter. Even though the cost per 100 declines, students will still pay more for additional cards.
