Evaluating and Choosing Cost Drivers

Learning Objective 5 Describe three criteria used to evaluate and choose cost drivers

• economically plausible relationships, goodness of fit, and significant effect of the cost driver on costs

Identifying cost drivers and estimating cost functions requires a good understanding of operations. Consider the costs to maintain and repair metal-cutting machines at Helix Corporation, a manufacturer of treadmills. Helix schedules repairs and maintenance during periods of low production to avoid having to take machines out of service when they are needed most. An analysis of the monthly data will then show high repair costs in months of low production and low repair costs in months of high production. Someone unfamiliar with operations might conclude that there is an inverse relationship between production and repair costs. The engineering link between units produced and repair costs, however, is usually clear-cut. Over time, there is a cause-and-effect relationship: the higher the level of production, the higher the repair costs. To estimate the relationship correctly, operating managers and analysts must recognize that repair costs tend to lag behind periods of high production and hence must use production of prior periods as the cost driver.

In other cases, choosing a cost driver is more subtle and difficult. Consider again indirect manufacturing labor costs at Elegant Rugs. Both number of machine-hours and number of direct manufacturing labor-hours are plausible cost drivers of indirect manufacturing labor costs. Managers are unsure which is the better cost driver. Exhibit 10-7 presents weekly data on indirect manufacturing labor costs and number of machine-hours for the most recent 12-week period from Exhibit 10-3, together with data on the number of direct manufacturing labor-hours for the same period.

What guidance do the different cost-estimation methods provide for choosing among cost drivers? The industrial engineering method relies on analyzing physical relationships between cost drivers and costs, which are difficult to specify in this case. The conference method and the account analysis method require managers to use their best subjective judgment to choose a cost driver and to estimate fixed and variable components of the cost function. Managers cannot use these methods to explore and test alternative cost drivers. The major advantage of quantitative methods, such as regression analysis, is that they are objective, so managers can use them to evaluate different cost drivers.

The cost analyst first estimates the following regression equation for the firm’s indirect manufacturing labor costs using number of direct manufacturing labor-hours as the independent variable:

Exhibit 10-8 shows the plot of the data points for number of direct manufacturing labor-hours and indirect manufacturing labor costs and the regression line that best fits the data. Recall that Exhibit 10-6 shows the corresponding graph when number of machine-hours is the cost driver. To decide which of the two cost drivers Elegant Rugs should choose, the analyst compares the machine-hour regression equation and the direct manufacturing labor-hour regression equation and uses three criteria to make this evaluation.

1. ECONOMIC PLAUSIBILITY. Both cost drivers are economically plausible. However, in the state-of-the-art, highly automated production environment at Elegant Rugs, managers familiar with the operations believe that indirect manufacturing labor costs such as machine maintenance costs are likely to be more closely related to the number of machine-hours used than the number of direct manufacturing labor-hours used.
2. GOODNESS OF FIT. Compare Exhibits 10-6 and 10-8. The vertical differences between actual costs and predicted costs are much smaller for the machine-hours regression than for the direct manufacturing labor-hours regression. The number of machine-hours used has a stronger relationship—or goodness of fit—with indirect manufacturing labor costs.
3. SIGNIFICANCE OF THE INDEPENDENT VARIABLE. Again compare Exhibits 10-6 and 10-8 (both of which have been drawn to roughly the same scale). The machine-hours regression line has a steep slope relative to the slope of the direct manufacturing labor-hours regression line. For the same (or more) scatter of observations about the line (goodness of fit), a flat or slightly sloped regression line indicates a weak relationship between the cost driver and costs. In our example, changes in the direct manufacturing labor-hours appear to have a small effect on the indirect manufacturing labor costs.

Based on this evaluation, managers at Elegant Rugs select number of machine-hours as the cost driver and use the cost function $y=\mathrm{\$}300.98+(\mathrm{\$}10.31\hspace{0.17em}\text{per\ machine}‐\text{hour}\times \text{Number\hspace{0.17em}of\hspace{0.17em}machine}‐\text{hours})$ to predict future indirect manufacturing labor costs.

Instructors and students who want to explore how regression analysis techniques can be used to choose among different cost drivers can go to Appendix: Regression Analysis and return to this point without any loss of continuity.

Why is choosing the correct cost driver to estimate indirect manufacturing labor costs important? Because identifying the wrong cost drivers or misestimating cost functions can lead to incorrect (and costly) decisions. Consider the following strategic decision at Elegant Rugs. The company is evaluating a new style of carpet that is similar to carpets it has manufactured in the past. It expects to sell 650 square feet of this carpet each week, requiring 72 machine-hours and 21 direct manufacturing labor-hours. Based on the machine-hour regression equation, Elegant Rugs would predict indirect manufacturing labor costs of Using direct manufacturing labor-hours as the cost driver, it would incorrectly predict costs of $\mathrm{\$}744.67+(\mathrm{\$}7.72\hspace{0.17em}\text{per\ labor}‐\text{hour}\times 21\hspace{0.17em}\text{labor}‐\text{hours})=\mathrm{\$}906.79\mathrm{.}$ If other incorrect cost drivers also underestimated other indirect costs, managers would conclude that the costs of manufacturing the new style of carpet are low. But actual indirect costs driven by the number of machine-hours used and other correct cost drivers would be higher. Failing to identify the proper cost drivers leads managers to believe the new style of carpet is more profitable than it actually is. If managers had correctly estimated costs and profitability, they may have decided not to introduce it.

Incorrectly estimating the cost function also affects cost management and cost control. Suppose number of direct manufacturing labor-hours was used as the cost driver, and actual indirect manufacturing labor costs for the new carpet were $990, higher than the predicted costs of $906.79. Managers would then feel compelled to cut costs. In fact, based on the appropriate machine-hour cost driver, actual costs are lower than the $1,043.30 predicted costs—a performance that management should seek to replicate, not change!

Cost Drivers and Activity-Based Costing

Activity-based costing (ABC) systems focus on individual activities, such as product design, machine setup, materials handling, distribution, and customer service, as the fundamental cost objects. Managers identify a cost driver for each activity. Consider materials-handling costs at Westronics, an electronic products manufacturer. Managers must decide whether number of loads moved or weight of loads moved is the cost driver of these costs. To do so, they apply methods described earlier in this chapter on data collected over a reasonably long period when any cause-and-effect relationship between the cost driver and costs is not masked by costs being fixed in the short run.

Managers apply all the methods described earlier to estimate cost relationships in ABC systems. The City of London police force uses input–output relationships (the industrial engineering method) to identify cost drivers and the cost of an activity. Using surveys of time required, officials determine costs associated with responding to house robberies, dealing with burglaries, and filling out police reports. The industrial engineering method is also used by U.S. government agencies, such as the U.S. Postal Service, to determine the cost of each post office transaction, and the U.S. Patent and Trademark Office, to identify the costs of each patent examination. Caterpillar uses the industrial engineering method to calculate the normalized cost of each activity in its manufacturing processes. Activity costs are then rolled up to the product level.

When choosing among methods to estimate cost functions, managers trade off level of detail, accuracy, feasibility, and costs. For example, to estimate the cost of opening a bank account or making a transfer payment, Bankinter in Spain uses work measurement methods, while Royal Bank of Canada uses advanced analytical techniques, including regression.

Increasingly, managers are using quantitative analysis to determine the cost drivers of activities. DHL Express, the international shipping company, recently switched from the conference method to performing in-depth quantitative analysis on its “big data” system. This system gives DHL the opportunity to do sophisticated analysis on large data sets at relatively low cost. A single, worldwide activity-based costing system shows the cost and profitability for every shipment in its network. Algorithms optimize the most profitable way to allocate shipments to its fleet of planes.

When estimating a cost function for a cost pool, managers must pay careful attention to the cost hierarchy. For example, if a cost is a batch-level cost such as setup cost, managers must only consider batch-level cost drivers like number of setup-hours. In some cases, the costs in a cost pool may have more than one cost driver from different levels of the cost hierarchy. The cost drivers for Elegant Rugs’ indirect manufacturing labor costs could be machine-hours and the number of production batches of carpet manufactured. Furthermore, it may be difficult to subdivide the indirect manufacturing labor costs into two cost pools and to measure the costs associated with each cost driver. In cases like these, companies use multiple regression, as we discuss in the chapter appendix, to estimate costs based on more than one independent variable.