The Portable MBA in Finance and Accounting, 3rd Edition

120 Understanding the Numbers

analyzed. To emphasize this, the cost function, Total Cost=(76%)Revenue
+$5 million, was used to generate the data set in Exhibit 3.12. A randomized
error term was then added to these data estimates, they were rounded to the
nearest quarter million, and then the high and low data points, July and Sep-
tember, were purposely changed. For instance, assume September was a very
busy month for Books “ ” Us because of the many college-student book or-
ders. This rush caused overtime and other disruptive cost behavior. Without
the analyst first adjusting the data point for this aberrant behavior, the results
are skewed. For databased techniques such as these, the adage “Garbage in,
garbage out” holds true. Before employing any of these techniques first ensure
that your data does truly ref lect the cost structure being studied.

R

EXHIBIT 3.14 Least-squares regression output (Books “R” Us data).

SUMMARY OUTPUT

Regression Statistics

Multiple R 99.1%
Rsquare 98.2%
Adjusted
Rsquare 98.0%
Standard
error 471.36
Obser vations 12

ANOVA

df SS MS F Significance F

Regression 1 119,835,495 119835495 539.363 4.956E-10
Residual 10 2,221,797 222179.69
Total 11 122,057,292

Coefficients

Intercept $2,733
Xvariable 1 90%

EXHIBIT 3.15 Databased cost structure estimates.

Variable Cost Fixed Cost

Percentage (in millions)

Visual fit 85 $4.0

High-low 98 1.725

Least squares 90 2.733