246 Chapter 5

E X H I B I T 5 A – 3  Y
The Concept of Least-Squares
Regression

Actual Y
Error Regression line

Cost
Estimated Y
Y = a + bX

X
Level of activity

Fortunately, Microsoft Excel can be used to estimate the fixed cost (intercept) and vari-
able cost per unit (slope) that minimize the sum of the squared errors. Excel also provides
a statistic called the R2, which is a measure of “goodness of fit.” The R2 tells us the per-
centage of the variation in the dependent variable (cost) that is explained by variation in
the independent variable (activity). The R2 varies from 0% to 100%, and the higher the
percentage, the better. 
As mentioned earlier, you should always plot the data in a scattergraph, but it is par-
ticularly important to check the data visually when the R2 is low. A quick look at the scat-
tergraph can reveal that there is little relation between the cost and the activity or that the
relation is something other than a simple straight line. In such cases, additional analysis
would be required.
Exhibit 5A–4 uses Excel to depict the Brentline Hospital data that we used earlier to
illustrate the high-low method. We’ll be using this same data set to illustrate how Excel
can be used to create a scattergraph plot and to calculate the intercept a, the slope b, and
the R2 using least-squares regression.6

E X H I B I T 5 A – 4 
The Least-Squares Regression
Worksheet for Brentline Hospital

6 
The authors wish to thank Don Schwartz, Professor of Accounting at National University, for provid-
ing suggestions that were instrumental in creating this appendix.
 Cost-Volume-Profit Relationships 247

EXHIBIT 5A–5
A Scattergraph Plot for Brentline
Hospital Using Microsoft Excel

To prepare a scattergraph plot in Excel, begin by highlighting the data in cells B4

through C10 (as shown in Exhibit 5A–4). From the Charts group within the Insert tab,
select the “Scatter” subgroup and then click on the choice that has no lines connect-
ing the data points. This should produce a scattergraph plot similar to the one shown
in Exhibit 5A–5. Notice that the number of patient-days is plotted on the X-axis and
the maintenance cost is plotted on the Y-axis.7 As we saw verified earlier in Exhibit 5A–1,
the data is approximately linear, so it makes sense to proceed with estimating a regres-
sion equation that minimizes the sum of the squared errors.
To determine the intercept a, the slope b, and the R2, begin by right clicking on
any data point in the scattergraph plot and selecting “Add Trendline.” This should pro-
duce the screen that is shown in Exhibit 5A–6. Notice that under “Trend/Regression
Type” you should select “Linear.” Similarly, under “Trendline Name” you should select
“Automatic.” Next to the word “Backward” you should input the lowest value for the
independent variable, which in this example is 5000 patient-days. Taking this particular
step instructs Excel to extend your fitted line until it intersects the Y-axis. Finally, you
should check the two boxes at the bottom of Exhibit 5A–6 that say “Display Equation
on chart” and “Display R-squared value on chart.”
Once you have established these settings, then click “Close.” As shown in
Exhibit 5A–7, this will automatically insert a line within the scattergraph plot that
minimizes the sum of the squared errors. It will also cause the estimated least-squares
regression equation and R2 to be inserted into your scattergraph plot. Instead of
depicting the results using the form Y = a + bX, Excel uses an equivalent form of
the equation depicted as Y = bX + a. In other words, Excel reverses the two terms
shown to the right of the equals sign. So, in Exhibit 5A–7, Excel shows a least-squares
regression equation of y = 0.7589x + 3,430.9. The slope b in this equation of $0.7589
represents the estimated variable maintenance cost per patient-day. The intercept a in
this equation of $3,430.90 (or approximately $3,431) represents the estimated fixed
monthly maintenance cost. Note that the R2 is approximately 0.90, which is quite good
and indicates that 90% of the variation in maintenance cost is explained by the variation
in patient-days.

7 
To insert labels for the X-axis and Y-axis, go to the Layout tab in Excel. Then, within the Labels
group, select Axis Titles.
248 Chapter 5

E X H I B I T 5 A – 6 
Trendline Options in Microsoft Excel

E X H I B I T 5 A – 7 
Brentline Hospital: Least-
Squares Regression Results
Using Microsoft Excel

Comparing the High-Low and Least-Squares

Regression Methods
The table below compares Brentline Hospital’s cost estimates using the high-low method
and the least-squares regression method:

High-Low Least-Squares
Method Regression Method
Variable cost estimate per patient-day������������� $0.800 $0.759
Fixed cost estimate per month��������������������������� $3,400 $3,431
 Cost-Volume-Profit Relationships 249

When Brentline uses the least-squares regression method to create a straight line that
minimizes the sum of the squared errors, it results in estimated fixed costs that are $31
higher than the amount derived using the high-low method. It also decreases the slope
of the straight line resulting in a lower variable cost estimate of $0.759 per patient-day
rather than $0.80 per patient-day as derived using the high-low method.

Glossary (Appendix 5A)

Account analysis  A method for analyzing cost behavior in which an account is classified as either
variable or fixed based on the analyst’s prior knowledge of how the cost in the account behaves.
(p. 242)
Dependent variable  A variable that responds to some causal factor; total cost is the dependent
variable, as represented by the letter Y, in the equation Y = a + bX. (p. 242)
Engineering approach  A detailed analysis of cost behavior based on an industrial engineer’s
evaluation of the inputs that are required to carry out a particular activity and of the prices of
those inputs. (p. 242)
High-low method  A method of separating a mixed cost into its fixed and variable elements by
analyzing the change in cost between the high and low activity levels. (p. 243)
Independent variable  A variable that acts as a causal factor; activity is the independent variable,
as represented by the letter X, in the equation Y = a + bX. (p. 242)
Least-squares regression method  A method of separating a mixed cost into its fixed and variable
elements by fitting a regression line that minimizes the sum of the squared errors. (p. 245)
Linear cost behavior  Cost behavior is said to be linear whenever a straight line is a reasonable
approximation for the relation between cost and activity. (p. 242)
R2  A measure of goodness of fit in least-squares regression analysis. It is the percentage of the varia-
tion in the dependent variable that is explained by variation in the independent variable. (p. 246)

Appendix 5A: Exercises and Problems

EXERCISE 5A–1 High-Low Method LO5–10
The Cheyenne Hotel in Big Sky, Montana, has accumulated records of the total electrical costs
of the hotel and the number of occupancy-days over the last year. An occupancy-day represents a
room rented for one day. The hotel’s business is highly seasonal, with peaks occurring during the
ski season and in the summer.

Month Occupancy-Days Electrical Costs

January �������������������������������������������������������� 1,736 $4,127
February�������������������������������������������������������� 1,904 $4,207
March������������������������������������������������������������ 2,356 $5,083
April �������������������������������������������������������������� 960 $2,857
May���������������������������������������������������������������� 360 $1,871
June�������������������������������������������������������������� 744 $2,696
July���������������������������������������������������������������� 2,108 $4,670
August ���������������������������������������������������������� 2,406 $5,148
September���������������������������������������������������� 840 $2,691
October�������������������������������������������������������� 124 $1,588
November���������������������������������������������������� 720 $2,454
December���������������������������������������������������� 1,364 $3,529

Required:
1. Using the high-low method, estimate the fixed cost of electricity per month and the variable
cost of electricity per occupancy-day. Round off the fixed cost to the nearest whole dollar and
the variable cost to the nearest whole cent.
2. What other factors in addition to occupancy-days are likely to affect the variation in electrical
costs from month to month?