Supplementary Problem 3: Cost Estimation – High-Low Method
As we’ve discussed, costs may exhibit varied cost behavior patterns. Cost
estimation is the process of determining how a particular cost behaves. Chapter 2 in your
text describes several methods of estimating costs. Some of these are relatively simple
(e.g., account classification) while others are much more costly and/or complex (e.g.,
motion and time studies, regression). This supplementary problem describes another
method of cost estimation – the high-low method.
In the high-low method, two data points are used to approximate fixed and
variable costs. The two data points that will be used are selected by finding the high and
low activity levels (NOT the cost levels) from the available data. Let us denote the high
activity level and associated cost as Ahigh and Chigh , respectively, and let us denote the low
activity level and associated cost as Alow and Clow , respectively. These two data points are
then used to compute fixed and variable costs as follows:
Variable cost per unit = Change in the cost of the two activity levels
Change in the activity level
= Chigh - Clow
Ahigh - Alow
This formula gives the variable cost per unit. Next, we need to compute the total variable
cost at either the high or low activity level. Multiplying the variable cost per unit by the
activity level at either the high or low point, we obtain the total variable cost. Finally, we
obtain the estimate of fixed cost by subtracting total variable cost from total cost. Note:
if you select the high activity level in the computation of total variable cost, you must use
the total cost corresponding to the high activity level in determining fixed cost. You now
have all the information necessary to develop a cost formula.
Note that this method is more objective than the account classification method
while only slightly more complex. However, only two data points are used to estimate
cost behavior, and to the extent that these data points are outliers, the cost estimation is
flawed. Additionally, since only two data points are used in the estimation procedure, any
information from all other data points between the two extreme activity levels is ignored.
Given this weakness, statistical techniques (e.g., regression) may be employed.
