HORN Ch10 SM

CHAPTER 10
DETERMINING HOW COSTS BEHAVE
10-1 What two assumptions are frequently made when estimating a cost function?
The two assumptions are

1. Variations in the level of a single activity (the cost driver) explain the variations in the
related total costs.
2. Cost behavior is approximated by a linear cost function within the relevant range. A
linear cost function is a cost function where, within the relevant range, the graph of total
costs versus the level of a single activity forms a straight line.
10-2 Describe three alternative linear cost functions.
Three alternative linear cost functions are

1. Variable cost function––a cost function in which total costs change in proportion to the
changes in the level of activity in the relevant range.
2. Fixed cost function––a cost function in which total costs do not change with changes in
the level of activity in the relevant range.
3. Mixed cost function––a cost function that has both variable and fixed elements. Total
costs change but not in proportion to the changes in the level of activity in the relevant
range.
10-3 What is the difference between a linear and a nonlinear cost function? Give an example
of each type of cost function.
A linear cost function is a cost function where, within the relevant range, the graph of total costs
versus the level of a single activity related to that cost is a straight line. An example of a linear
cost function is a cost function for use of a videoconferencing line where the terms are a fixed
charge of $10,000 per year plus a $2 per minute charge for line use. A nonlinear cost function is
a cost function where, within the relevant range, the graph of total costs versus the level of a
single activity related to that cost is not a straight line. Examples include economies of scale in
advertising where an agency can double the number of advertisements for less than twice the
costs, step-cost functions, and learning-curve-based costs.
10-4 “High correlation between two variables means that one is the cause and the other is the
effect.” Do you agree? Explain.
No. High correlation merely indicates that the two variables move together in the data examined.
It is essential also to consider economic plausibility before making inferences about cause and
effect. Without any economic plausibility for a relationship, it is less likely that a high level of
correlation observed in one set of data will be similarly found in other sets of data.
10-5 Name four approaches to estimating a cost function.
Four approaches to estimating a cost function are
10-1
1. Industrial engineering method.
2. Conference method.
3. Account analysis method.
4. Quantitative analysis of current or past cost relationships.
10-6 Describe the conference method for estimating a cost function. What are two advantages
of this method?
The conference method estimates cost functions on the basis of analysis and opinions about costs
and their drivers gathered from various departments of a company (purchasing, process
engineering, manufacturing, employee relations, etc.). Advantages of the conference method
include
1. The speed with which cost estimates can be developed.
2. The pooling of knowledge from experts across functional areas.
3. The improved credibility of the cost function to all personnel.
10-7 Describe the account analysis method for estimating a cost function.
The account analysis method estimates cost functions by classifying cost accounts in the
subsidiary ledger as variable, fixed, or mixed with respect to the identified level of activity.
Typically, managers use qualitative, rather than quantitative, analysis when making these cost-
classification decisions.
10-8 List the six steps in estimating a cost function on the basis of an analysis of a past cost
relationship. Which step is typically the most difficult for the cost analyst?
The six steps are

1. Choose the dependent variable (the variable to be predicted, which is some type of cost).
2. Identify the independent variable or cost driver.
3. Collect data on the dependent variable and the cost driver.
4. Plot the data.
5. Estimate the cost function.
6. Evaluate the cost driver of the estimated cost function.
Step 3 typically is the most difficult for a cost analyst.
10-9 When using the high-low method, should you base the high and low observations on the
dependent variable or on the cost driver?
Causality in a cost function runs from the cost driver to the dependent variable. Thus, choosing
the highest observation and the lowest observation of the cost driver is appropriate in the high-
low method.
10-10 Describe three criteria for evaluating cost functions and choosing cost drivers.
Three criteria important when choosing among alternative cost functions are
1. Economic plausibility.
10-2
2. Goodness of fit.
3. Slope of the regression line.
10-11 Define learning curve. Outline two models that can be used when incorporating learning
into the estimation of cost functions.
A learning curve is a function that measures how labor-hours per unit decline as units of
production increase because workers are learning and becoming better at their jobs. Two models
used to capture different forms of learning are
1. Cumulative average-time learning model. The cumulative average time per unit declines
by a constant percentage each time the cumulative quantity of units produced doubles.
2. Incremental unit-time learning model. The incremental time needed to produce the last
unit declines by a constant percentage each time the cumulative quantity of units
produced doubles.
10-12 Discuss four frequently encountered problems when collecting cost data on variables
included in a cost function.
Frequently encountered problems when collecting cost data on variables included in a cost
function are
1. The time period used to measure the dependent variable is not properly matched with the
time period used to measure the cost driver(s).
2. Fixed costs are allocated as if they are variable.
3. Data are either not available for all observations or are not uniformly reliable.
4. Extreme values of observations occur.
5. A homogeneous relationship between the individual cost items in the dependent variable
cost pool and the cost driver(s) does not exist.
6. The relationship between the cost and the cost driver is not stationary.
7. Inflation has occurred in a dependent variable, a cost driver, or both.
10-13 What are the four key assumptions examined in specification analysis in the case of
simple regression?
Four key assumptions examined in specification analysis are

1. Linearity of relationship between the dependent variable and the independent variable
within the relevant range.
2. Constant variance of residuals for all values of the independent variable.
3. Independence of residuals.
4. Normal distribution of residuals.
10-14 “All the independent variables in a cost function estimated with regression analysis are
cost drivers.” Do you agree? Explain.
No. A cost driver is any factor whose change causes a change in the total cost of a related cost
object. A cause-and-effect relationship underlies selection of a cost driver. Some users of
regression analysis include numerous independent variables in a regression model in an attempt
10-3
to maximize goodness of fit, irrespective of the economic plausibility of the independent
variables included. Some of the independent variables included may not be cost drivers.
10-15 “Multicollinearity exists when the dependent variable and the independent variable are
highly correlated.” Do you agree? Explain.
No. Multicollinearity exists when two or more independent variables are highly correlated with
each other.
10-16 HL Co. uses the high-low method to derive a total cost formula. Using a range of units
produced from 1,500 to 7,500, and a range of total costs from $21,000 to $45,000, producing
2,000 units will cost HL:
a. $8,000 b. $12,000
c. $23,000 d. $29,000
SOLUTION
Choice "c" is correct. The high-low method is used to estimate both fixed and variable costs, and
can then be applied to determine a total cost formula that is used to estimate total costs for any
level of production.
The difference between the total costs ($45,000 − $21,000) is divided by the difference in units
(7,500 – 1,500) to derive a variable cost per unit of $4 ($24,000 / 6,000). Using either end of the
range, fixed costs can then be estimated. Using total costs of $45,000 for 7,500 units, with
variable costs at $4 per unit, $45,000 − 7,500($4) = $15,000 of fixed costs.
The total cost formula for HL will be equal to: $15,000 + [$4.00 × # units]. 2,000 units will
produce a total cost of: $15,000 + [$4.00 × 2,000] = $23,000.
Choice "a" is incorrect. This calculation fails to account for the fixed costs of $15,000.
Choice "b" is incorrect. This calculation incorrectly assumes that because 7,500 units cost
$45,000 (or $6 overall per unit), that 2,000 units would cost $12,000 ($6 per unit).
Choice "d" is incorrect. This calculation incorrectly applies a variable cost of $7 per unit rather
than $4.
10-17 A firm uses simple linear regression to forecast the costs for its main product line. If
fixed costs are equal to $235,000 and variable costs are $10 per unit, how many units does it
need to sell at $15 per unit to make a $300,000 profit?
10-4
a. 21,400 b. 47,000
c. 60,000 d. 107,000
SOLUTION
Choice "d" is correct. The regression equation set up will be: y (total costs) = $235,000 + $10x,
with x representing volume. In order to make a $300,000 profit, sales ($15x) − costs must equal
$300,000. So the full set up will be: $15x − ($235,000 + $10x) = $300,000. Solving for x, $5x =
$535,000, or 107,000 units. At 107,000 units, sales will total $1,605,000 and costs will total
$1,305,000 for a profit of $300,000.
Choice "a" is incorrect. This choice represents a calculation error where the $15 sale price and
the $10 variable cost are added together and divided into $535,000.
Choice "b" is incorrect. 47,000 is the number of units that is required in order to breakeven.
Choice "c" is incorrect. These are the number of units above breakeven that the company must
sell in order to make a $300,000 profit.
10-18 In regression analysis, the coefficient of determination:
a. Is used to determine the proportion of the total variation in the dependent variable (y)
explained by the independent variable (X).
b. Ranges between negative one and positive one.
c. Is used to determine the expected value of the net income based on the regression line.
d. Becomes smaller as the fit of the regression line improves.
SOLUTION
Choice "a" is correct. This is the definition of the coefficient of determination. It is the square of
the coefficient of correlation. The higher the coefficient of determination, the greater the
proportion of the total variation in y that is explained by the variation in x. The higher it is, the
better is the fit of the regression line.
Choice "b" is incorrect. It ranges between 0 and 1. Remember, the coefficient of determination is
the square of the coefficient of correlation. Because it is a number squared, it will be positive.
Choice "c" is incorrect. This is not a use of the coefficient of determination.
Choice "d" is incorrect. It becomes larger as the fit of the regression line improves.
10-19 A regression equation is set up, where the dependent variable is total costs and the
independent variable is production. A correlation coefficient of 0.70 implies that:
a. The coefficient of determination is negative.
10-5
b. The level of production explains 49% of the variation in total costs
c. There is a slightly inverse relationship between production and total costs.
d. A correlation coefficient of 1.30 would produce a regression line with better fit to the data.
SOLUTION
Choice "b" is correct. A correlation coefficient (used to measure the strength in the linear
relationship between independent and dependent variables) of 0.70 implies that the coefficient of
determination is 0.49. A coefficient of determination of 0.49 equates to the independent variable
(level of production) explaining 49 percent of the variation in the dependent variable (total
costs).
Choice "a" is incorrect. The coefficient of determination will always be a number between 0 and
1.
Choice "c" is incorrect. A positive correlation coefficient implies a direct relationship between
the two variables.
Choice "d" is incorrect. The correlation coefficient can only be between -1 and 1.
10-20 What would be the approximate value of the coefficient of correlation between
advertising and sales where a company advertises aggressively as an alternative to temporary
worker layoffs and cuts off advertising when incoming jobs are on backorder?
a. 1.0 b. 0
c. –1.0 d. –100
SOLUTION
Choice "c" is correct. The coefficient of correlation measures the strength and direction of the
relationship between two variables. Since the company increases advertising when sales are low
and decreases advertising when sales are high, the movement is in directly opposite directions
and the coefficient would be close to - 1.0.
Choice "a" is incorrect. A coefficient of correlation of 1.0 would imply that both variables move
in the same direction at approximately the same rate. An increase in advertising when sales are
increasing would be characteristic of a correlation of coefficient of 1.0.
Choice "b" is incorrect. A coefficient of correlation of 0 would imply that there is no relationship
between advertising and sales. There is an inverse relationship between advertising and sales.
Choice "d" is incorrect. A relationship exists between advertising and sales. According to the
facts of the question, the relationship is an inverse relationship. The coefficient of correlation is
expressed as a range between 1.0 and +1.0.
10-21 Estimating a cost function. The controller of the Javier Company is preparing the budget
10-6
for 2018 and needs to estimate a cost function for delivery costs. Information regarding delivery
costs incurred in the prior two months are:
Month Miles Driven Delivery Costs

August 12,000 $10,000
September 17,000 $13,000
Required:
1. Estimate the cost function for delivery.
2. Can the constant in the cost function be used as an estimate of fixed delivery cost per month?
Explain.
SOLUTION
(10 min.) Estimating a cost function.
Difference in delivery costs

1. Slope coefficient =
Difference in miles driven
$13,000 − $10,000
=
17,000 − 12,000
$3,000
= = $0.60 per mile
5,000
Constant = Total cost – (Slope coefficient × Quantity of cost driver)
= $10,000 – ($0.60 × 12,000) = $2,800
= $13,000 – ($0.60 × 17,000) = $2,800
The cost function based on the two observations is
Delivery costs = $2,800 + $0.60 × Miles driven
2. The cost function in requirement 1 is an estimate of how costs behave within the relevant
range, not at cost levels outside the relevant range. If there are no months with zero miles driven
represented in the delivery cost account in the general ledger, data in that account cannot be used
to estimate the fixed costs at the zero miles driven level. Rather, the constant component of the
cost function provides the best available starting point for a straight line that approximates how a
cost behaves within the relevant range.
10-7
10-22 Identifying variable-, fixed-, and mixed-cost functions. The Sunrise Corporation
operates car rental agencies at more than 20 airports. Customers can choose from one of three
contracts for car rentals of one day or less:
¡ Contract 1: $45 for the day

¡ Contract 2: $25 for the day plus $0.30 per mile traveled
¡ Contract 3: $1.50 per mile traveled
Required:
1. Plot separate graphs for each of the three contracts, with costs on the vertical axis and miles
traveled on the horizontal axis.
2. Express each contract as a linear cost function of the form y = a + bX .
3. Identify each contract as a variable-, fixed-, or mixed-cost function.
SOLUTION
(15 min.) Identifying variable-, fixed-, and mixed-cost functions.
1. See Solution Exhibit 10-22.
2. Contract 1: y = $45
Contract 2: y = $25 + $0.30X
Contract 3: y = $1.50X
where X is the number of miles traveled in the day.
3. Contract Cost Function

1 Fixed
2 Mixed
3 Variable
SOLUTION EXHIBIT 10-22

Plots of Car Rental Contracts Offered by Sunrise Corp.
10-8
Contract 1: Fixed Costs
$100
$80
Car Rental Costs
$60
$40
$20
$0
0 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150
Miles Traveled Per Day
Contract 2: Mixed Costs

$100
$80
Car Rental Costs
$60
$40
$20
$0
0 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150
10-9
Contract 3: Variable Costs
$250
$200
Car Rental Costs
$150
$100
$50
$0
0 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150
10-23 Various cost-behavior patterns. (CPA, adapted).
The vertical axes of the graphs below represent total cost, and the horizontal axes represent units
produced during a calendar year. In each case, the zero point of dollars and production is at the
intersection of the two axes.
Required:
Select the graph that matches the numbered manufacturing cost data (requirements 1–9). Indicate
by letter which graph best fits the situation or item described. The graphs may be used more than
once.
1. Annual depreciation of equipment, where the amount of depreciation charged is computed by
10-10
the machine-hours method.
2. Electricity bill—a flat fixed charge, plus a variable cost after a certain number of kilowatt-
hours are used, in which the quantity of kilowatt-hours used varies proportionately with
quantity of units produced.
3. City water bill, which is computed as follows:
First 1,000,000 gallons or less $1,000 flat fee

Next 10,000 gallons $0.003 per gallon used
and so on and so on
The gallons of water used vary proportionately with the quantity of production output.
4. Cost of direct materials, where direct material cost per unit produced decreases with each
pound of material used (for example, if 1 pound is used, the cost is $10; if 2 pounds are used,
the cost is $19.98; if 3 pounds are used, the cost is $29.94), with a minimum cost per unit of
$9.20.
5. Annual depreciation of equipment, where the amount is computed by the straight-line
method. When the depreciation schedule was prepared, it was anticipated that the
obsolescence factor would be greater than the wear-and-tear factor.
6. Rent on a manufacturing plant donated by the city, where the agreement calls for a fixed-fee
payment unless 200,000 labor-hours are worked, in which case no rent is paid.
7. Salaries of repair personnel, where one person is needed for every 1,000 machine-hours or
less (that is, 0 to 1,000 hours requires one person, 1,001 to 2,000 hours requires two people,
and so on).
8. Cost of direct materials used (assume no quantity discounts).
9. Rent on a manufacturing plant donated by the county, where the agreement calls for rent of
$100,000 to be reduced by $1 for each direct manufacturing labor-hour worked in excess of
200,000 hours, but a minimum rental fee of $20,000 must be paid.
SOLUTION
(20 min.) Various cost-behavior patterns.

1. K
2. B
3. G
4. J Note that A is incorrect because, although the cost per pound eventually equals a
constant at $9.20, the total dollars of cost increases linearly from that point
onward.
5. I The total costs will be the same regardless of the volume level.
6. L
7. F This is a classic step-cost function.
8. K
9. C
10-11
10-24 Matching graphs with descriptions of cost and revenue behavior. (D. Green, adapted)
Given here are a number of graphs.
Required:
The horizontal axis of each graph represents the units produced over the year, and the vertical
axis represents total cost or revenues.
Indicate by number which graph best fits the situation or item described (a–h). Some graphs
may be used more than once; some may not apply to any of the situations.
a. Direct material costs

b. Supervisors’ salaries for one shift and two shifts
c. A cost–volume–profit graph
d. Mixed costs—for example, car rental fixed charge plus a rate per mile driven
e. Depreciation of plant, computed on a straight-line basis
f. Data supporting the use of a variable-cost rate, such as manufacturing labor cost of $14 per unit
produced
g. Incentive bonus plan that pays managers $0.10 for every unit produced above some level of
production
h. Interest expense on $2 million borrowed at a fixed rate of interest
SOLUTION
(30 min.) Matching graphs with descriptions of cost and revenue behavior.
a. (1)
b. (6) A step-cost function.
c. (9)
d. (2)
e. (8)
10-12
f. (10) It is data plotted on a scatter diagram, showing a linear variable cost function with
constant variance of residuals. The constant variance of residuals implies that
there is a uniform dispersion of the data points about the regression line.
g. (3)
h. (8)
10-25 Account analysis, high-low. Stein Corporation wants to find an equation to estimate
some of their monthly operating costs for the operating budget for 2018. The following cost and
other data were gathered for 2017:
Maintenance Machine Health Number of Shipping Units

Month Costs Hours Insurance Employees Costs Shipped
January $4,500 165 $8,600 68 $25,776 7,160
February $4,452 120 $8,600 75 $29,664 8,240
March $4,600 230 $8,600 92 $28,674 7,965
April $4,850 318 $8,600 105 $23,058 6,405
May $5,166 460 $8,600 89 $21,294 5,915
June $4,760 280 $8,600 87 $33,282 9,245
July $4,910 340 $8,600 93 $31,428 8,730
August $4,960 360 $8,600 88 $30,294 8,415
September $5,070 420 $8,600 95 $25,110 6,975
October $5,250 495 $8,600 102 $25,866 7,185
November $5,271 510 $8,600 97 $20,124 5,590
December $4,760 275 $8,600 94 $34,596 9,610
Required:
1. Which of the preceding costs is variable? Fixed? Mixed? Explain.
2. Using the high-low method, determine the cost function for each cost.
3. Combine the preceding information to get a monthly operating cost function for the Stein
Corporation.
4. Next month, Stein expects to use 400 machine hours, have 80 employees, and ship 9,000
units. Estimate the total operating cost for the month.
SOLUTION
(20 min.) Account analysis, high-low
1. The maintenance cost is a mixed cost because the cost neither remains constant in total nor
remains constant per unit.
10-13
The health insurance cost is fixed because, although the number of employees varies from month
to month, the total cost remains constant at $8,600.
The shipping cost is variable because, in each month, the cost divided by the number of units
shipped equals a constant $3.60. The definition of a variable cost is one that remains constant per
unit.
2. The month with the highest number of machine hours is November, with 510 machine hours
and $5,271 of cost. The month with the lowest is February, with 120 machine hours and $4,452
in cost. The difference in cost ($5,271 – $4,452), divided by the difference in machine hours
(510 – 120) equals $2.10 per machine hour of variable maintenance cost. Inserted into the cost
formula for November:
$5,271 = a fixed cost + ($2.10 × number of machine hours used)

$5,271 = a + ($2.10 × 510)
$5,271 = a + $1,071
a = $4,200 monthly fixed maintenance cost
Therefore, Stein’s cost formula for monthly maintenance cost is:

y = $4,200 + ($2.10 × number of machine hours used)
3. The shipping rate is $3.60 per unit shipped

The maintenance cost is $4,200 + ($2.10 per machine hour used)
The health insurance cost is $8,600.
Adding them together we get:
Fixed operating costs = $4,200 (maintenance) + $8,600 (health insurance) = $12,800
Monthly Operating Cost = $12,800 + ($3.60 per unit shipped) + ($2.10 per machine hour used)
4. Estimated operating cost for November:

$12,800 + ($3.60 × 9,000 units shipped) + ($2.10 × 400 machine hours)
= $12,800 + $32,400 + $840
= $46,040
10-26 Account analysis method. Gower, Inc., a manufacturer of plastic products, reports the
following manufacturing costs and account analysis classification for the year ended December
31, 2017.
Account Classification Amount

Direct materials All variable $300,000
Direct manufacturing labor All variable 225,000
Power All variable 37,500
10-14
Account Classification Amount
Supervision labor 20% variable 56,250
Materials-handling labor 50% variable 60,000
Maintenance labor 40% variable 75,000
Depreciation 0% variable 95,000
Rent, property taxes, and administration 0% variable 100,000
Gower, Inc., produced 75,000 units of product in 2017. Gower’s management is estimating costs
for 2018 on the basis of 2017 numbers. The following additional information is available for
2018.
a. Direct materials prices in 2018 are expected to increase by 5% compared with 2017.
b. Under the terms of the labor contract, direct manufacturing labor wage rates are expected to
increase by 10% in 2018 compared with 2017.
c. Power rates and wage rates for supervision, materials handling, and maintenance are not
expected to change from 2017 to 2018.
d. Depreciation costs are expected to increase by 5%, and rent, property taxes, and
administration costs are expected to increase by 7%.
e. Gower expects to manufacture and sell 80,000 units in 2018.
Required:
1. Prepare a schedule of variable, fixed, and total manufacturing costs for each account category
in 2018. Estimate total manufacturing costs for 2018.
2. Calculate Gower’s total manufacturing cost per unit in 2017, and estimate total
manufacturing cost per unit in 2018.
3. How can you obtain better estimates of fixed and variable costs? Why would these better
estimates be useful to Gower?
SOLUTION
(30 min.) Account analysis method.
1. Manufacturing cost classification for 2017:

% of
Total Costs
Total That is Variable Fixed Variable
Costs Variable Costs Costs Cost per Unit
Account (1) (2) (3) = (1) × (2) (4) = (1) – (3) (5) = (3) ÷ 75,000
10-15
Direct materials $300,000 100% $300,000 $ 0 $4.00
Direct manufacturing labor 225,000 100 225,000 0 3.00
Power 37,500 100 37,500 0 0.50
Supervision labor 56,250 20 11,250 45,000 0.15
Materials-handling labor 60,000 50 30,000 30,000 0.40
Maintenance labor 75,000 40 30,000 45,000 0.40
Depreciation 95,000 0 0 95,000 0
Rent, property taxes, admin 100,000 0 0 100,000 0
Total $948,750 $633,750 $315,000 $8.45
Total manufacturing cost for 2017 = $948,750
Variable costs in 2018:

Unit
Variable Increase in
Cost per Variable Variable Cost
Unit for Percentage Cost per Unit Total Variable
2017 Increase per Unit for 2018 Costs for 2018
Account (6) (7) (8) = (6) × (7) (9) = (6) + (8) (10) = (9) × 80,000
Direct materials $4.00 5% $0.20 $4.20 $336,000

Direct manufacturing labor 3.00 10 0.30 3.30 264,000
Power 0.50 0 0 0.50 40,000
Supervision labor 0.15 0 0 0.15 12,000
Materials-handling labor 0.40 0 0 0.40 32,000
Maintenance labor 0.40 0 0 0.40 32,000
Depreciation 0 0 0 0 0
Rent, property taxes, admin. 0 0 0 0 0
Total $8.45 $0.50 $8.95 $716,000
Fixed and total costs in 2018:

Dollar
Fixed Increase in Fixed Costs Variable Total
Costs Percentage Fixed Costs for 2018 Costs for Costs
for 2018 Increase (13) = (14) = 2018 (16) =
Account (11) (12) (11) × (12) (11) + (13) (15) (14) + (15)
Direct materials $ 0 0% $ 0 $ 0 $336,000 $ 336,000

Direct manufacturing labor 0 0 0 0 264,000 264,000
Power 0 0 0 0 40,000 40,000
Supervision labor 45,000 0 0 45,000 12,000 57,000
Materials-handling labor 30,000 0 0 30,000 32,000 62,000
Maintenance labor 45,000 0 0 45,000 32,000 77,000
Depreciation 95,000 5 4,750 99,750 0 99,750
Rent, property taxes, admin. 100,000 7 7,000 107,000 0 107,000
Total $315,000 $11,750 $326,750 $716,000 $1,042,750
Total manufacturing costs for 2018 = $1,042,750
10-16
$948,750
2. Total cost per unit, 2017 = = $12.65
75,000
$1,042,750
Total cost per unit, 2018 = = $13.03
80,000
3. Cost classification into variable and fixed costs is based on qualitative, rather than
quantitative, analysis. How good the classifications are depends on the knowledge of individual
managers who classify the costs. Gower may want to undertake quantitative analysis of costs,
using regression analysis on time-series or cross-sectional data to better estimate the fixed and
variable components of costs. Better knowledge of fixed and variable costs will help Gower to
better price his products, to know when he is getting a positive contribution margin, and to better
manage costs.
10-27 Estimating a cost function, high-low method. Reisen Travel offers helicopter service
from suburban towns to John F. Kennedy International Airport in New York City. Each of its 10
helicopters makes between 1,000 and 2,000 round-trips per year. The records indicate that a
helicopter that has made 1,000 round-trips in the year incurs an average operating cost of $350
per round-trip, and one that has made 2,000 round-trips in the year incurs an average operating
cost of $300 per round-trip.
Required:
1. Using the high-low method, estimate the linear relationship y = a + bX , where y is the total
annual operating cost of a helicopter and X is the number of round-trips it makes to JFK
airport during the year.
2. Give examples of costs that would be included in a and in b.
3. If Reisen Travel expects each helicopter to make, on average, 1,200 round-trips in the
coming year, what should its estimated operating budget for the helicopter fleet be?
SOLUTION
(15–20 min.) Estimating a cost function, high-low method.
1. The key point to note is that the problem provides high-low values of X (annual round
trips made by a helicopter) and Y ÷ X (the operating cost per round trip). We first need to
calculate the annual operating cost Y (as in column (3) below), and then use those values to
estimate the function using the high-low method.
Cost Driver: Operating Annual

Annual Round- Cost per Operating
Trips (X) Round-Trip Cost (Y)
(1) (2) (3) = (1) × (2)
Highest observation of cost driver 2,000 $300 $600,000
Lowest observation of cost driver 1,000 $350 $350,000
Difference 1,000 $250,000
10-17
Slope coefficient = $250,000 ÷ 1,000 = $250 per round-trip
Constant = $600,000 – ($250 × 2,000) = $100,000
The estimated relationship is Y = $100,000 + $250 X; where Y is the annual operating cost of a
helicopter and X represents the number of round trips it makes annually.
2. The constant a (estimated as $100,000) represents the fixed costs of operating a

helicopter, irrespective of the number of round trips it makes. This would include items such as
insurance, registration, depreciation on the aircraft, and any fixed component of pilot and crew
salaries. The coefficient b (estimated as $250 per round-trip) represents the variable cost of each
round trip—costs that are incurred only when a helicopter actually flies a round trip. The
coefficient b may include costs such as landing fees, fuel, refreshments, baggage handling, and
any regulatory fees paid on a per-flight basis.
3. If each helicopter is, on average, expected to make 1,200 round trips a year, we can use
the estimated relationship to calculate the expected annual operating cost per helicopter:
Y = $100,000 + $250 X
X = 1,200
Y = $100,000 + $250 × 1,200 = $100,000 + $300,000 = $400,000
With 10 helicopters in its fleet, Reisen’s estimated operating budget is 10 × $400,000 = $4,000,000.
10-28 Estimating a cost function, high-low method. Lacy Dallas is examining customer-
service costs in the southern region of Camilla Products. Camilla Products has more than 200
separate electrical products that are sold with a 6-month guarantee of full repair or replacement
with a new product. When a product is returned by a customer, a service report is prepared. This
service report includes details of the problem and the time and cost of resolving the problem.
Weekly data for the most recent 8-week period are as follows:
Week Customer-Service Department Costs Number of Service Reports

1 $13,300 185
2 20,500 285
3 12,000 120
4 18,500 360
5 14,900 275
6 21,600 440
7 16,500 350
8 21,300 315
Required:
1. Plot the relationship between customer-service costs and number of service reports. Is the
relationship economically plausible?
10-18
2. Use the high-low method to compute the cost function relating customer-service costs to the
number of service reports.
3. What variables, in addition to number of service reports, might be cost drivers of weekly
customer-service costs of Camilla Products?
SOLUTION
(20 min.) Estimating a cost function, high-low method.
1. See Solution Exhibit 10-28. There is a positive relationship between the number of
service reports (a cost driver) and the customer-service department costs. This relationship is
economically plausible.
2. Number of Customer-Service
Service Reports Department Costs
Highest observation of cost driver 440 $21,600
Lowest observation of cost driver 120 12,000
Difference 320 $ 9,600
Customer-service department costs = a + b (number of service reports)
$9,600
Slope coefficient (b) = = $30 per service report
320
Constant (a) = $21,600 – ($30 × 440) = $8,400
= $12,000 – ($30 × 120) = $8,400
Customer-service
= $8,400 + $30 (number of service reports)
department costs
3. Other possible cost drivers of customer-service department costs are:

a. Number of products replaced with a new product (and the dollar value of the new
products charged to the customer-service department).
b. Number of products repaired and the time and cost of repairs.
Plot of Number of Service Reports versus Customer-Service Dept. Costs for Camilla Products
10-19
$24,000
Customer-Service Department Costs

$22,000
$20,000
$18,000
$16,000
$14,000
$12,000
$10,000
100 150 200 250 300 350 400 450 500
Number of Service Reports
10-29 Linear cost approximation. Dr. Young, of Young and Associates, LLP, is examining
how overhead costs behave as a function of monthly physician contact hours billed to patients.
The historical data are as follows:
Total Overhead Costs Physician Contact Hours Billed to Patients

$ 90,000 150
105,000 200
111,000 250
125,000 300
137,000 350
150,000 400
Required:
1. Compute the linear cost function, relating total overhead costs to physician contact hours,
using the representative observations of 200 and 300 hours. Plot the linear cost function.
Does the constant component of the cost function represent the fixed overhead costs of
Young and Associates? Why?
2. What would be the predicted total overhead costs for (a) 150 hours and (b) 400 hours using
the cost function estimated in requirement 1? Plot the predicted costs and actual costs for 150
and 400 hours.
3. Dr. Young had a chance to do some school physicals that would have boosted physician
contact hours billed to patients from 200 to 250 hours. Suppose Dr. Young, guided by the
linear cost function, rejected this job because it would have brought a total increase in
contribution margin of $9,000, before deducting the predicted increase in total overhead cost,
$10,000. What is the total contribution margin actually forgone?
SOLUTION
10-20
(30–40 min.) Linear cost approximation.
Difference in overhead costs

1. Slope coefficient (b) =
Difference in contact hours billed
= ($125,000 - $105,000)/(300 – 200)

= $200
Constant (a) = $125,000 – ($200 × 300)

= $65,000
Cost function = $65,000 + ($200 × contact hours billed)
The linear cost function is plotted in Solution Exhibit 10-29.
No, the constant component of the cost function does not represent the fixed overhead cost of
Young and Associates. The relevant range of physician contact hours billed is from 150 to 400.
The constant component provides the best available starting point for a straight line that
approximates how a cost behaves within the relevant range.
2. A comparison at various levels of contact hours billed follows. The linear cost function is
based on the formula of $65,000 per month plus $200 per physician contact hours billed.
Total overhead cost behavior:

Month 1 Month 2 Month 3 Month 4 Month 5 Month 6
Professional labor-hours 150 200 250 300 350 400
Actual total overhead costs $90,000 $105,000 $111,000 $125,000 $137,000 $150,000
Linear approximation 95,000 105,000 115,000 125,000 135,000 145,000
Actual minus linear
Approximation $(5,000) $ 0 $ (4,000) $ 0 $ 2,000 $ 5,000
The data are shown in Solution Exhibit 10-29. The linear cost function overstates costs by
$5,000 at the 150-hour level and understates costs by $5,000 at the 400-hour level.
3. Based on Based on Linear

Actual Cost Function
Contribution before deducting incremental overhead $9,000 $9,000
Incremental overhead 6,000 10,000
Contribution after incremental overhead $ 3,000 $ (1,000)
The total contribution margin actually forgone is $3,000.

Linear Cost Function Plot of Physician Contact Hours Billed
and Total Overhead Costs for Young and Associates
10-21
$160,000
$150,000
Total Overhead Costs

$140,000
$130,000
$120,000
$110,000
$100,000
$90,000
$80,000
100 150 200 250 300 350 400 450 500
Physician Contact Hours Billed
10-30 Cost-volume-profit and regression analysis. Relling Corporation manufactures a drink

bottle, model CL24. During 2017, Relling produced 210,000 bottles at a total cost of $808,500.
Kraff Corporation has offered to supply as many bottles as Relling wants at a cost of $3.75 per
bottle. Relling anticipates needing 225,000 bottles each year for the next few years.
Required:
1. a. What is the average cost of manufacturing a drink bottle in 2017? How does it compare to
Kraff’s offer?
b. Can Relling use the answer in requirement 1a to determine the cost of manufacturing
225,000 drink bottles? Explain.
2. Relling’s cost analyst uses annual data from past years to estimate the following regression
equation with total manufacturing costs of the drink bottle as the dependent variable and
drink bottles produced as the independent variable:
y = $445,000 + $1.75 X
During the years used to estimate the regression equation, the production of bottles varied from
200,000 to 235,000. Using this equation, estimate how much it would cost Relling to
manufacture 225,000 drink bottles. How much more or less costly is it to manufacture the
bottles than to acquire them from Kraff?
3. What other information would you need to be confident that the equation in requirement 2
accurately predicts the cost of manufacturing drink bottles?
SOLUTION
(20 min.) Cost-volume-profit and regression analysis.
Total manufacturing costs

1a. Average cost of manufacturing =
Number of drink bottles
10-22
$808,500
= = $3.85 per bottle
210,000
This cost is higher than the $3.75 per bottle that Kraff has quoted.
1b. Rellings cannot take the average manufacturing cost in 2017 of $3.85 per bottle and
multiply it by 225,000 drink bottles to determine the total cost of manufacturing 225,000 drink
bottles. The reason is that some of the $808,500 (or equivalently the $3.85 cost per bottle) are
fixed costs and some are variable costs. Without distinguishing fixed from variable costs,
Rellings cannot determine the cost of manufacturing 225,000 drink bottles. For example, if all
costs are fixed, the manufacturing costs of 225,000 bottles will continue to be $808,500. If,
however, all costs are variable, the cost of manufacturing 225,000 bottles would be $3.85 ×
225,000 = $866,250. If some costs are fixed and some are variable, the cost of manufacturing
225,000 bottles will be somewhere between $808,500 and $866,250.
Some students could argue that another reason for not being able to determine the cost of
manufacturing 225,000 drink bottles is that not all costs are output unit-level costs. If some costs
are, for example, batch-level costs, more information would be needed on the number of batches
in which the 225,000 drink bottles would be produced, in order to determine the cost of
manufacturing 225,000 drink bottles.
Expected cost to make

2. = $445,000 + ($1.75 × 225,000)
225,000 drink bottles
= $445,000 + $393,750 = $838,750
Purchasing drink bottles from Kraff will cost $3.75 × 225,000 = $843,750. Hence, it will
cost Rellings $843,750 − $838,750 = $5,000 more to purchase the bottles from Kraff rather than
manufacture them in-house.
3. Rellings would need to consider several factors before being confident that the equation
in requirement 2 accurately predicts the cost of manufacturing drink bottles.
a. Is the relationship between total manufacturing costs and quantity of drink bottles
economically plausible? For example, is the quantity of bottles made the only cost
driver or are there other cost-drivers (for example batch-level costs of setups,
production-orders or material handling) that affect manufacturing costs?
b. How good is the goodness of fit? That is, how well does the estimated line fit the
data?
c. Is the relationship between the number of drink bottles produced and total
manufacturing costs linear?
d. Does the slope of the regression line indicate that a strong relationship exists between
manufacturing costs and the number of drink bottles produced?
e. Are there any data problems such as, for example, errors in measuring costs, trends in
prices of materials, labor or overheads that might affect variable or fixed costs over
10-23
time, extreme values of observations, or a nonstationary relationship over time
between total manufacturing costs and the quantity of bottles produced?
f. How is inflation expected to affect costs?
g. Will Kraff supply high-quality drink bottles on time?
h. What is the length of the agreement with Kraff? Short-term or Long-term?
10-31 Regression analysis, service company. (CMA, adapted) Linda Olson owns a
professional character business in a large metropolitan area. She hires local college students to
play these characters at children’s parties and other events. Linda provides balloons, cupcakes,
and punch. For a standard party the cost on a per-person basis is as follows:
Balloons, cupcakes, and punch $ 7

Labor (0.25 hour × $20 per hour) 5
Overhead (0.25 hour × $40 per hour 10
Total cost per person $22
Linda is quite certain about the estimates of the materials and labor costs, but is not as
comfortable with the overhead estimate. The overhead estimate was based on the actual data for
the past 9 months, which are presented here. These data indicate that overhead costs vary with
the direct labor-hours used. The $40 estimate was determined by dividing total overhead costs
for the 9 months by total labor-hours.
Month Labor-Hours Overhead

Costs
April 1,400 $ 65,000
May 1,800 71,000
June 2,100 73,000
July 2,200 76,000
August 1,650 67,000
September 1,725 68,000
October 1,500 66,500
November 1,200 60,000
December 1,900 72,500
Total 15,475 $619,000
Linda has recently become aware of regression analysis. She estimated the following regression
equation with overhead costs as the dependent variable and labor-hours as the independent
variable:
y = $43,563 + $14.66 X
10-24
Required:
1. Plot the relationship between overhead costs and labor-hours. Draw the regression line and
evaluate it using the criteria of economic plausibility, goodness of fit, and slope of the
regression line.
2. Using data from the regression analysis, what is the variable cost per person for a standard
party?
3. Linda Olson has been asked to prepare a bid for a 20-child birthday party to be given next
month. Determine the minimum bid price that Linda would be willing to submit to recoup
variable costs.
SOLUTION
(25 min.)Regression analysis, service company.
1. Solution Exhibit 10-31 plots the relationship between labor-hours and overhead costs and
shows the regression line.
y = $43,563 + $14.66 X
Economic plausibility. Labor-hours appears to be an economically plausible driver of

overhead costs for the character company. Overhead costs such as scheduling, hiring and training
of workers, and managing the workforce are largely incurred to support labor.
Goodness of fit. The vertical differences between actual and predicted costs are extremely
small, indicating a very good fit. The good fit indicates a strong relationship between the labor-
hour cost driver and overhead costs.
Slope of regression line. The regression line has a reasonably steep slope from left to
right. Given the small scatter of the observations around the line, the positive slope indicates that,
on average, overhead costs increase as labor-hours increase.
2. The regression analysis indicates that, within the relevant range of 1,200 to 2,200 labor-
hours, the variable cost per person for a birthday party equals:
Balloons, cupcakes and punch $ 7.00

Labor (0.25 hrs. × $20 per hour) 5.00
Variable overhead (0.25 hrs. × $14.66 per labor-hour) 3.67
Total variable cost per person $15.67
3. To earn a positive contribution margin, the minimum bid for a 20-child birthday party
would be any amount greater than $313.40. This amount is calculated by multiplying the variable
cost per child of $15.67 by the 20 children. At a price above the variable cost of $313.40, Linda
Olson will be earning a contribution margin toward coverage of her fixed costs.
Of course, Linda Olson will consider other factors in developing her bid including (a) an
analysis of the competition––vigorous competition will limit Linda’s ability to obtain a higher
10-25
price (b) a determination of whether or not her bid will set a precedent for lower prices––overall,
the prices Linda Olson charges should generate enough contribution to cover fixed costs and earn
a reasonable profit, and (c) a judgment of how representative past historical data (used in the
regression analysis) is about future costs.

Regression Line of Overhead Costs on Labor-Hours for Linda Olson’s Character Business
80,000
75,000 y = 14.664x + 43563

R² = 0.95514
Overhead Costs
70,000
65,000
60,000
55,000
50,000
1,000 1,200 1,400 1,600 1,800 2,000 2,200 2,400
Labor-Hours
10-32 High-low, regression. May Blackwell is the new manager of the materials storeroom for
Clayton Manufacturing. May has been asked to estimate future monthly purchase costs for part
#696, used in two of Clayton’s products. May has purchase cost and quantity data for the past 9
months as follows:
Month Cost of Purchase Quantity Purchased

January $12,675 2,710 parts
February 13,000 2,810
March 17,653 4,153
April 15,825 3,756
May 13,125 2,912
June 13,814 3,387
July 15,300 3,622
August 10,233 2,298
September 14,950 3,562
Estimated monthly purchases for this part based on expected demand of the two products for the
rest of the year are as follows:
10-26
Month Purchase Quantity Expected
October 3,340 parts
November 3,710
December 3,040
Required:
1. The computer in May’s office is down, and May has been asked to immediately provide an
equation to estimate the future purchase cost for part #696. May grabs a calculator and uses
the high-low method to estimate a cost equation. What equation does she get?
2. Using the equation from requirement 1, calculate the future expected purchase costs for each
of the last 3 months of the year.
3. After a few hours May’s computer is fixed. May uses the first 9 months of data and
regression analysis to estimate the relationship between the quantity purchased and purchase
costs of part #696. The regression line May obtains is as follows:
y = $2,582.6 + 3.54 X
Evaluate the regression line using the criteria of economic plausibility, goodness of fit, and
significance of the independent variable. Compare the regression equation to the equation
based on the high-low method. Which is a better fit? Why?
4. Use the regression results to calculate the expected purchase costs for October, November,
and December. Compare the expected purchase costs to the expected purchase costs
calculated using the high-low method in requirement 2. Comment on your results.
SOLUTION
(25 min.) High-low, regression
1. May will pick the highest point of activity, 4,153 parts (March) at $17,653 of cost, and
the lowest point of activity, 2,298 parts (August) at $10,233.
Cost driver:
Quantity Purchased Cost
Highest observation of cost driver 4,153 $17,653
Lowest observation of cost driver 2,298 10,233
Difference 1,855 $ 7,420
Purchase costs = a + b × Quantity purchased
Slope Coefficient = $7,420/1,855 = $4 per part
Constant (a) = $17,653 ─ ($4 × 4,153) = $1,041
The equation May gets is:
10-27
Purchase costs = $1,041 + ($4 × Quantity purchased)
2. Using the equation above, the expected purchase costs for each month will be:
Purchase
Quantity Expected
Month Expected Formula cost
October 3,340 parts y = $1,041 + ($4 × 3,340) $14,401
November 3,710 y = $1,041 + ($4 × 3,710) 15,881
December 3,040 y = $1,041 + ($4 × 3,040) 13,201
3. Economic Plausibility: Clearly, the cost of purchasing a part is associated with the
quantity purchased.
Goodness of Fit: As seen in Solution Exhibit 10-32, the regression line fits the data well. The
vertical distance between the regression line and observations is small. An r-squared value of
close to 0.96 indicates that almost 98% of the change in cost can be explained by the change in
quantity purchased.
Significance of the Independent Variable: The relatively steep slope of the regression line
suggests that the quantity purchased is correlated with purchasing cost for part #696.
Clayton Manufacturing Purchase Costs for Part #696

$20,000
$18,000 y = 3.5376x + 2582.6

R² = 0.95828
Cost of Purchase
$16,000
$14,000
$12,000
$10,000
$8,000
2,000 2,500 3,000 3,500 4,000 4,500
Quantity Purchased
According to the regression, May’s original estimate of fixed cost is too low given all the data
points. The original slope is too steep, but only by 46 cents. So, the variable rate is lower but
the fixed cost is higher for the regression line than for the high-low cost equation.
10-28
The regression is the more accurate estimate because it uses all available data (all nine data
points) while the high-low method only relies on two data points and may therefore miss some
important information contained in the other data.
4. Using the regression equation, the purchase costs for each month will be:
Purchase
Quantity
Month Expected Formula Expected cost
October 3,340 parts y = $2,582.60 + ($3.54 × 3,340) $14,406.20
November 3,710 y = $2,582.60 + ($3.54 × 3,710) 15,716.00
December 3,040 y = $2,582.60 + ($3.54 × 3,040) 13,344.20
Although the two equations are different in both fixed element and variable rate, within the
relevant range they give similar expected costs. This implies that the high and low points of the
data are a reasonable representation of the total set of points within the relevant range.
10-33 Learning curve, cumulative average-time learning model. Northern Defense

manufactures radar systems. It has just completed the manufacture of its first newly designed
system, RS-32. Manufacturing data for the RS-32 follow:
Required:
Calculate the total variable costs of producing 2, 4, and 8 units.
SOLUTION
(20 min.) Learning curve, cumulative average-time learning model.
The direct manufacturing labor-hours (DMLH) required to produce the first 2, 4, and 8 units
given the assumption of a cumulative average-time learning curve of 85%, is as follows:
85% Learning Curve
Cumulative Cumulative Cumulative
10-29
Number Average Time Total Time:
of Units (X) per Unit (y): Labor Hours Labor-Hours
(1) (2) (3) = (1) × (2)
1 4,400 4,400
2 3,740 = (4,400 × 0.85) 7,480
4 3,179 = (3,740 × 0.85) 12,716
8 2,702 = (3,179 × 0.85) 21,617
Alternatively, to compute the values in column (2) we could use the formula
y = aXb
where a = 4,400, X = 2, 4, or 8, and b = – 0.234465, which gives

when X = 2, y = 4,400× 2– 0.234465 = 7,480
when X = 4, y = 4,400× 4– 0.234465 = 12,716
when X = 8, y = 4,400× 8– 0.234465 = 21,617
Variable Costs of Producing

2 Units 4 Units 8 Units
Direct materials $84,000 × 2; 4; 8 $168,000 $336,000 $672,000
Direct manufacturing labor
$27 × 7,480; 12,716; 21,617 201,960 343,332 583,664
Variable manufacturing overhead
$13 × 7,480; 12,716; 21,617 97,240 165,308 281,024
Total variable costs $467,200 $844,640 $1,536,688
10-34 Learning curve, incremental unit-time learning model. Assume the same information
for Northern Defense as in Exercise 10-33, except that Northern Defense uses an 85%
incremental unit-time learning model as a basis for predicting direct manufacturing labor-hours.
(An 85% learning curve means b = –0.234465.)
Required:
1. Calculate the total variable costs of producing 2, 3, and 4 units.
2. If you solved Exercise 10-33, compare your cost predictions in the two exercises for 2 and 4
units. Why are the predictions different? How should Northern Defense decide which model
it should use?
SOLUTION
(20 min.) Learning curve, incremental unit-time learning model.
1. The direct manufacturing labor-hours (DMLH) required to produce the first 2, 3, and 4
units, given the assumption of an incremental unit-time learning curve of 85%, is as follows:
85% Learning Curve
Cumulative Individual Unit Time for Xth Cumulative Total Time:
Number of Units (X) Unit (y): Labor Hours Labor-Hours
10-30
(1) (2) (3)
1 4,400 4,400
2 3,740 = (4,400 × 0.85) 8,140
3 3,401 11,541
4 3,179 = (3,740 × 0.85) 14,720
Values in column (2) are calculated using the formula y = aXb where a = 4,400, X = 2, 3,
or 4, and b = – 0.234465, which gives
when X = 2, y = 4,400 × 2– 0.234465 = 3,740
when X = 3, y = 4,400 × 3– 0.234465 = 3,401
when X = 4, y = 4,400 × 4– 0.234465 = 3,179
Variable Costs of Producing

2 Units 3 Units 4 Units
Direct materials $84,000 × 2; 3; 4 $168,000 $ 252,000 $ 336,000
Direct manufacturing labor
$27 × 8,140; 11,541; 14,720 219,780 311,607 397,440
Variable manufacturing overhead
$13 × 8,140; 11,541; 14,720 105,820 150,033 191,360
Total variable costs $493,600 $713,640 $924,800
2. Variable Costs of
Producing
2 Units 4 Units
Incremental unit-time learning model (from requirement 1) $493,600 $924,800
Cumulative average-time learning model (from Exercise 10-34) 467,200 844,640
Difference $ 26,400 $ 80,160
Total variable costs for manufacturing 2 and 4 units are lower under the cumulative
average-time learning curve relative to the incremental unit-time learning curve. Direct
manufacturing labor-hours required to make additional units decline more slowly in the
incremental unit-time learning curve relative to the cumulative average-time learning curve when
the same 85% factor is used for both curves. The reason is that, in the incremental unit-time
learning curve, as the number of units double only the last unit produced has a cost of 85% of the
initial cost. In the cumulative average-time learning model, doubling the number of units causes
the average cost of all the units produced (not just the last unit) to be 85% of the initial cost.
10-35 High-low method. Wayne Mueller, financial analyst at CELL Corporation, is examining
the behavior of quarterly utility costs for budgeting purposes. Mueller collects the following data
on machine-hours worked and utility costs for the past 8 quarters:
Quarter Machine-Hours Utility Costs

1 120,000 $215,000
2 75,000 150,000
10-31
3 110,000 200,000
4 150,000 270,000
5 90,000 170,000
6 140,000 250,000
7 130,000 225,000
8 100,000 195,000
Required:
1. Estimate the cost function for the quarterly data using the high-low method.
2. Plot and comment on the estimated cost function.
3. Mueller anticipates that CELL will operate machines for 125,000 hours in quarter 9.
Calculate the predicted utility costs in quarter 9 using the cost function estimated in
requirement 1.
SOLUTION
(25 min.) High-low method.
1. Machine-Hours Utility Costs
Highest observation of cost driver 150,000 $270,000

Lowest observation of cost driver 75,000 150,000
Difference 75,000 $120,000
Utility costs = a + b × Machine-hours
$120,000
Slope coefficient (b) = = $1.60 per machine-hour
75,000
Constant (a) = $270,000 – ($1.60 × 150,000)
= $270,000 – $240,000 = $30,000
or Constant (a) = $150,000 – ($1.60 × 75,000)
= $150,000 – $120,000 = $30,000
Utility costs = $30,000 + ($1.60 × Machine-hours)
2.
Plot and High-Low Line of Utility Costs as a Function of Machine-Hours
10-32
$280,000
$260,000
$240,000
$220,000
Utility Costs
$200,000
$180,000
$160,000
$140,000
$120,000
$100,000
60,000 80,000 100,000 120,000 140,000 160,000
Machine-Hours
Solution Exhibit 10-35 presents the high-low line.
Economic plausibility. The cost function shows a positive economically plausible relationship
between machine-hours and utility costs. There is a clear-cut engineering relationship of higher
machine-hours and utility costs.
Goodness of fit. The high-low line appears to “fit” the data well. The vertical differences
between the actual and predicted costs appear to be quite small.
Slope of high-low line. The slope of the line appears to be reasonably steep indicating that, on
average, utility costs in a quarter vary with machine-hours used.
3. Using the cost function estimated in 1, predicted utility costs would be:
$30,000 + ($1.60 × 125,000 hours) = $230,000.
Mueller should budget $230,000 in quarter 9 because the relationship between machine-
hours and utility costs in Solution Exhibit 10-35 is economically plausible, has an excellent
goodness of fit, and indicates that an increase in machine-hours in a quarter causes utility costs to
increase in the quarter.
10-36 High-low method and regression analysis. Market Thyme, a cooperative of organic
family-owned farms, has recently started a fresh produce club to provide support to the group’s
member farms and to promote the benefits of eating organic, locally produced food. Families pay
a seasonal membership fee of $100 and place their orders a week in advance for a price of $40
per order. In turn, Market Thyme delivers fresh-picked seasonal local produce to several
neighborhood distribution points. Five hundred families joined the club for the first season, but
the number of orders varied from week to week.
Tom Diehl has run the produce club for the first season. Tom is now a farmer but remembers
a few things about cost analysis from college. In planning for next year, he wants to know how
many orders will be needed each week for the club to break even, but first he must estimate the
club’s fixed and variable costs. He has collected the following data over the club’s first season of
operation:
10-33
Week Number of Orders per Week Weekly Total Costs
1 415 $26,900
2 435 27,200
3 285 24,700
4 325 25,200
5 450 27,995
6 360 25,900
7 420 27,000
8 460 28,315
9 380 26,425
10 350 25,750
Required:
1. Plot the relationship between number of orders per week and weekly total costs.
2. Estimate the cost equation using the high-low method, and draw this line on your graph.
3. Tom uses his computer to calculate the following regression formula:
Weekly total costs = $18,791 + ($19.97 × Number of orders per week)
Draw the regression line on your graph. Use your graph to evaluate the regression line using
the criteria of economic plausibility, goodness of fit, and significance of the independent
variable. Is the cost function estimated using the high-low method a close approximation of
the cost function estimated using the regression method? Explain briefly.
4. Did Market Thyme break even this season? Remember that each of the families paid a
seasonal membership fee of $100.
5. Assume that 500 families join the club next year and that prices and costs do not change.
How many orders, on average, must Market Thyme receive each of 10 weeks next season to
break even?
SOLUTION
(30min.) High-low method and regression analysis.
1. See Solution Exhibit 10-36.

10-34
$29,000
$28,500
$28,000
Weekly Total Costs

$27,500 High-Low Line
$27,000
$26,500
$26,000
$25,500 Regression Line
$25,000
$24,500
$24,000
250 300 350 400 450 500
Number of Weekly Orders
2.
Number of Weekly
Orders per week Total Costs
Highest observation of cost driver (Week 8) 460 $28,315
Lowest observation of cost driver (Week 3) 285 24,700
Difference 175 $ 3,615
Weekly total costs = a + b (number of orders per week)
Slope coefficient (b) = $3,615/175=$20.66 per order
Constant (a) = $28,315 – ($20.66 × 460) = $18,811.14

= $24,700 – ($20.66 × 285) = $18,811.14
Weekly total costs = $18,811.14 + $20.66 × (Number of Orders per week)
See high-low line in Solution Exhibit 10-36.
3. Solution Exhibit 10-36 presents the regression line:
Weekly total costs = $18,791 + $19.97 × (Number of Orders per week)
Economic Plausibility. The cost function shows a positive economically plausible relationship
between number of orders per week and weekly total costs. Number of orders is a plausible cost
driver of total weekly costs.
Goodness of fit. The regression line appears to fit the data well. The vertical differences
between the actual costs and the regression line appear to be quite small.
10-35
Significance of independent variable. The regression line has a steep positive slope and
increases by $19.97 for each additional order. Because the slope is not flat, there is a strong
relationship between number of orders and total weekly costs.
The regression line is the more accurate estimate of the relationship between number of orders
and total weekly costs because it uses all available data points while the high-low method relies
only on two data points and may therefore miss some information contained in the other data
points. In addition, because both the low and high data points lie off the regression line, the
high-low method predicts a higher amount of fixed costs as well as a steeper slope (higher
amount of variable cost per order).
4. Profit =
Total weekly revenues + Total seasonal membership fees – Total weekly costs
(Total number of orders × $40) + (500 × $100) – $265,385 =
(3,880 × $40) + (500 × $100) – $265,385 =
$155,200 + $50,000 – $265,385 = ($60,185).
No, the club did not make a profit.
5. Let the average number of weekly orders be denoted by AWO. We want to find the value
of AWO for which Market Thyme will achieve zero profit.
Using the format in requirement 4, we want:
Profit = [AWO × 10 weeks × $40] + (500 × $100) – [$18,791 + ($19.97 × AWO)] × 10 weeks
$0 = ($400 AWO) + $50,000 – $187,910 – $199.70 AWO
$200.30 AWO = $137,910
AWO = $137,910 ÷ $200.30 = 688.52
So, Market Thyme will have to get at least orders on average each week in order to break even
next year.
10-37 High-low method; regression analysis. (CIMA, adapted) Catherine McCarthy, sales
manager of Baxter Arenas, is checking to see if there is any relationship between promotional
costs and ticket revenues at the sports stadium. She obtains the following data for the past 9
months:
Month Ticket Revenues Promotional Costs

April $200,000 $52,000
May 270,000 65,000
June 320,000 80,000
July 480,000 90,000
10-36
August 430,000 100,000
September 450,000 110,000
October 540,000 120,000
November 670,000 180,000
December 751,000 197,000
She estimates the following regression equation:
Ticket revenues = $65,583 + ($3.54 × Promotional costs)
Required:
1. Plot the relationship between promotional costs and ticket revenues. Also draw the regression
line and evaluate it using the criteria of economic plausibility, goodness of fit, and slope of
the regression line.
2. Use the high-low method to compute the function relating promotional costs and revenues.
3. Using (a) the regression equation and (b) the high-low equation, what is the increase in
revenues for each $10,000 spent on promotional costs within the relevant range? Which
method should Catherine use to predict the effect of promotional costs on ticket revenues?
Explain briefly.
SOLUTION
(30−40 min.) High-low method, regression analysis.
1. Solution Exhibit 10-37 presents the plots of promotional costs on revenues.

Plot and Regression Line of Promotional Costs on Ticket Revenues
10-37
840,000
y = 3.542x + 65583
740,000 R² = 0.93476
640,000
Ticket Revenues 540,000

440,000
340,000
240,000
140,000
40,000
50,000 70,000 90,000 110,000 130,000 150,000 170,000 190,000
Promotional Costs
Solution Exhibit 10-37 also shows the regression line of advertising costs on revenues.
We evaluate the estimated regression equation using the criteria of economic plausibility,
goodness of fit, and slope of the regression line.
Economic plausibility. Promotional costs appear to be a plausible cost driver of ticket revenues.
Sports teams and sporting events frequently use promotions to entice numbers of people to
attend.
Goodness of fit. The vertical differences between actual and predicted revenues appears to be
reasonably small. This indicates that promotional costs are related to ticket revenues.
Slope of regression line. The slope of the regression line appears to be relatively steep. Given the
small scatter of the observations around the line, the steep slope indicates that, on average, ticket
revenues increase with spending on promotions.
2. The high-low method would estimate the cost function as follows:
Promotional Ticket
Costs Revenues
Highest observation of revenue driver $ 52,000 $200,000
Lowest observation of revenue driver 197,000 751,000
Difference $145,000 $551,000
Revenues = a + (b × promotional costs)
$551,000
Slope coefficient (b) = = 3.80
$145,000
10-38
Constant (a) = $200,000 − ($52,000 × 3.80)
= $200,000 − $197,600 = $2,400
or Constant (a) = $751,000 − ($197,000 × 3.80)
= $751,000 − $748,600 = $2,400
Revenues = $2,400 + (3.80 × Promotional costs)
3. The increase in revenues for each $1,000 spent on advertising within the relevant range is
a. Using the regression equation, 3.542 × $10,000 = $35,420

b. Using the high-low equation, 3.80 × $10,000 = $38,000
The high-low equation does moderately well in estimating the relationship between
promotional costs and revenues; it overestimates the impact of marginal changes in promotional
expenses on ticket revenues. McCarthy should use the regression equation because it uses
information from all observations. The high-low method, on the other hand, relies only on the
observations that have the highest and lowest values of the cost driver and these observations are
generally not representative of all the data.
10-38 Regression, activity-based costing, choosing cost drivers. Sleep Late, a large hotel
chain, has been using activity-based costing to determine the cost of a night’s stay at their hotels.
One of the activities, “Inspection,” occurs after a customer has checked out of a hotel room.
Sleep Late inspects every 10th room and has been using “number of rooms inspected” as the cost
driver for inspection costs. A significant component of inspection costs is the cost of the supplies
used in each inspection.
Mary Adams, the chief inspector, is wondering whether inspection labor-hours might be a
better cost driver for inspection costs. Mary gathers information for weekly inspection costs,
rooms inspected, and inspection labor-hours as follows:
Week Rooms Inspected Inspection Labor-Hours Inspection Costs

1 254 66 $1,740
2 322 110 2,500
3 335 82 2,250
4 431 123 2,800
5 198 48 1,400
6 239 62 1,690
7 252 108 1,720
8 325 127 2,200
10-39
Mary runs regressions on each of the possible cost drivers and estimates these cost functions:
Inspection Costs = $193.19 + ($6.26 × Number of rooms inspected)
Inspection Costs = $944.66 + ($12.04 × Inspection labor-hours)
Required:
1. Explain why rooms inspected and inspection labor-hours are plausible cost drivers of
inspection costs.
2. Plot the data and regression line for rooms inspected and inspection costs. Plot the data and
regression line for inspection labor-hours and inspection costs. Which cost driver of
inspection costs would you choose? Explain.
3. Mary expects inspectors to inspect 300 rooms and work for 105 hours next week. Using the
cost driver you chose in requirement 2, what amount of inspection costs should Mary
budget? Explain any implications of Mary choosing the cost driver you did not choose in
requirement 2 to budget inspection costs.
SOLUTION
(30 min.) Regression, activity-based costing, choosing cost drivers.
1. Both number of rooms inspected and inspection labor-hours are plausible cost drivers for
inspection costs. The number of rooms inspected is likely related to the materials and supplies
that are used in each room, which is a significant component of inspection costs. Inspection
labor-hours are a plausible cost driver if labor hours vary per room inspected, because costs
would be a function of how much time the inspectors spend on each unit. This is particularly
true if the inspectors are paid a wage, and if they spend more time inspecting certain types of
rooms (e.g., suites) if they are larger or need to be more closely inspected because of the higher
prices being charged for them and the corresponding expectations of the guests.
2. Solution Exhibit 10-38 presents (a) the plots and regression line for number of rooms
inspected versus inspection costs and (b) the plots and regression line for inspection labor-hours
and inspection costs.
SOLUTION EXHIBIT 10-38A

Plot and Regression Line for Rooms Inspected versus Inspection Costs for Sleep Late
10-40
Sleep Late
$3,500
$3,000 y = 6.2625x + 193.19
R² = 0.93624
$2,500
Inspection costs
$2,000
$1,500
$1,000
$500
$0
200 250 300 350 400 450
Rooms inspected
SOLUTION EXHIBIT 10-38B

Plot and Regression Line for Inspection Labor-Hours and Inspection Costs for Sleep Late
Sleep Late
$3,000
y = 12.042x + 944.66
$2,500
R² = 0.583
Inspection costs
$2,000
$1,500
$1,000
$500
$0
40 60 80 100 120
Inspection labor-hours
Goodness of Fit. As you can see from the two graphs, the regression line based on number of
rooms inspected better fits the data (has smaller vertical distances from the points to the line)
than the regression line based on inspection labor-hours. The activity of inspection appears to be
more closely linearly related to the number of rooms inspected than to inspection labor-hours.
Hence number of rooms inspected is a better cost driver. This is probably because the number of
rooms inspected is closely related to the supplies used for inspection, and the variation due to
more or less time spent inspecting does not affect inspection costs (e.g., the inspectors may be
paid a salary and so higher labor-hours do not translate to higher costs).
10-41
Significance of independent variable. It is hard to visually compare the slopes because the
graphs are not the same size, but both graphs have steep positive slopes indicating a strong
relationship between number of rooms inspected and inspection costs, and inspection labor-hours
and inspection costs. Indeed, if labor-hours per inspection do not vary much, number of rooms
inspected and inspection labor-hours will be closely related. Overall, it is the significant cost of
supplies that is driven by the number of rooms inspected (not the inspection labor-hours spent on
inspection) that makes rooms inspected the preferred cost driver.
3. At 105 inspection labor hours and 300 rooms inspected:
Inspection costs using rooms inspected = $193.19 + ($6.26 × 300) = $2,071.19
Inspection costs using inspection labor-hours = $944.86 + ($12.04 × 105) = $2,209.06
If Mary uses inspection-labor-hours she will estimate inspection costs to be $2,209.06, $137.87
($2,209.06 ─ $2,071.19) higher than if she had used number of rooms inspected. If actual costs
equaled, say, $2,150, Mary would conclude that Sleep Late has performed efficiently in its
inspection activity because actual inspection costs would be lower than budgeted amounts. In
fact, based on the more accurate cost function, actual costs of $2,150 exceeded the budgeted
amount of $2,071.19. Mary should find ways to improve inspection efficiency rather than
mistakenly conclude that the inspection activity has been performing well.
10-39 Interpreting regression results. Spirit Freightways is a leader in transporting agricultural

products in the western provinces of Canada. Reese Brown, a financial analyst at Spirit
Freightways, is studying the behavior of transportation costs for budgeting purposes. Transportation
costs at Spirit are of two types: (a) operating costs (such as labor and fuel) and (b) maintenance
costs (primarily overhaul of vehicles).
Brown gathers monthly data on each type of cost, as well as the total freight miles traveled
by Spirit vehicles in each month. The data collected are shown below (all in thousands):
Month Operating Costs Maintenance Costs Freight Miles

January $ 942 $ 974 1,710
February 1,008 776 2,655
March 1,218 686 2,705
April 1,380 694 4,220
May 1,484 588 4,660
June 1,548 422 4,455
July 1,568 352 4,435
August 1,972 420 4,990
September 1,190 564 2,990
October 1,302 788 2,610
November 962 762 2,240
10-42
December 772 1,028 1,490
Required:
1. Conduct a regression using the monthly data of operating costs on freight miles. You should
obtain the following result:
Regression: Operating costs = a + (b × Number of freight miles)
Variable Coefficient Standard Error t-Value

Constant $445.76 $112.97 3.95
Independent variable: No. of freight miles $ 0.26 $ 0.03 7.83
2
r = 0.86; Durbin-Watson statistic = 2.18
2. Plot the data and regression line for the above estimation. Evaluate the regression using the
criteria of economic plausibility, goodness of fit, and slope of the regression line.
3. Brown expects Spirit to generate, on average, 3,600 freight miles each month next year. How
much in operating costs should Brown budget for next year?
4. Name three variables, other than freight miles, that Brown might expect to be important cost
drivers for Spirit’s operating costs.
5. Brown next conducts a regression using the monthly data of maintenance costs on freight
miles. Verify that she obtained the following result:
Regression: Maintenance costs = a + (b × Number of freight miles)

Constant $1,170.57 $91.07 12.85
Independent variable: No. of freight miles $ –0.15 $ 0.03 –5.83
2
6. Provide a reasoned explanation for the observed sign on the cost driver variable in the
maintenance cost regression. What alternative data or alternative regression specifications
would you like to use to better capture the above relationship?
SOLUTION
(15-20min.) Interpreting regression results, matching time periods.
1. Here is the regression data for monthly operating costs as a function of the total freight
miles travelled by Sprit vehicles:
SUMMARY OUTPUT
Regression Statistics
Multiple R 0.927299101
10-43
R Square 0.859883623
Adjusted R
Square 0.845871986
Standard Error 132.0816002
Observations 12
ANOVA
Significance
df SS MS F F
Regression 1 1070620.18 1070620.18 61.37 0.00
Residual 10 174455.49 17445.55
Total 11 1245075.67
Standard Upper
Coefficients Error t Stat P-value Lower 95% 95%
Intercept 445.76 112.97 3.95 0.00 194.04 697.48
X Variable 1 0.26 0.03 7.83 0.00 0.18 0.33
2. The chart below presents the data and the estimated regression line for the relationship
between monthly operating costs and freight miles traveled by Spirit Freightways.
Economic A positive relationship between freight miles traveled and monthly

plausibility operating costs is economically plausible since increased levels of
economic activity should lead to the consumption of greater amounts of
labor, fuel and other operating expenses.
Goodness of fit r2 = 86%, Adjusted r2 = 85%

Standard error of regression = 132.08
Excellent fit; there is indisputable evidence of a linear relationship between
the dependent and independent variables. The distances between the
estimated line and the actual data points are small, other than at the highest
10-44
level of activity recorded during the year.
Significance of The t-value of 7.83 for freight miles traveled output units is significant at
Independent the 0.05 and 0.01 levels.
Variables
3. If Brown expects Spirit to generate an average of 3,600 miles each month next year, the best
estimate of operating costs is given by:
Monthly operating costs = $445.76 + ($0.26) × (3,600 miles) = $1,381.76.
Annual operating costs = ($1,381.76) × 12 = $16,581.12.
4. Three variables, other than freight miles, that Brown might expect to be important cost drivers
for Spirit’s operating costs are: input prices (fuel prices and wage rates), mix of agricultural
output carried (weight, volume, value), and route mix and conditions (weather, flat versus
mountainous terrain, short-haul versus long-haul carriage).
5. Here is the regression data for monthly maintenance costs as a function of the total freight
miles travelled by Sprit vehicles:
SUMMARY OUTPUT
R Square 0.77241808
Adjusted R
Square 0.74965989
Observations 12
ANOVA
Significance
df SS MS F F
Regression 1 384747.37 384747.37 33.94 0.00
Residual 10 113360.30 11336.03
Total 11 498107.67
Standard Upper
Intercept 1170.57 91.07 12.85 0.00 967.66 1373.48
X Variable 1 -0.15 0.03 -5.83 0.00 -0.21 -0.09
The data and regression estimate are provided in the chart below:
10-45
6. At first glance, the regression result in requirement 5 is surprising and economically-
implausible. In the regression, the coefficient on freight miles traveled has a negative sign. This
implies that the greater the number of freight miles (i.e., the more activity Spirit carries out), the
smaller are the maintenance costs; specifically, it suggests that each extra freight mile reduces
maintenance costs by $0.14 (recall that all data are in thousands). Clearly, this estimated
relationship is not economically credible. However, one would think that freight miles should
have some impact on fleet maintenance costs.
The logic behind the estimated regression becomes clearer once one realizes that maintenance
costs have a discretionary component to them, especially in terms of timing. Spirit’s peak months
of work transporting agricultural products in western Canada occur in late spring and summer
(the period from April through August). It is likely that Spirit is simply choosing to defer
maintenance to those months when its vehicles are not in use, thereby creating a negative
relationship between monthly activity and maintenance costs. The causality also goes the other
way – if vehicles are in the shop for maintenance, they are clearly not on the road generating
freight miles. A third reason is that vehicles might need to be serviced at greater frequency
during the winter months because of the wear and tear that comes from driving on icy terrain and
in poor weather conditions.
Possible alternative specifications that would better capture the link between Spirit’s activity
levels and the spending on maintenance are to estimate the relationship using annual data over a
period of several years, to look at spending on corrective rather than preventive maintenance, or
to look at the relation using lags (i.e., freight miles traveled in a period against the spending on
maintenance done in a subsequent period in order to service the vehicles).
10-40 Cost estimation, cumulative average-time learning curve. The Pacific Boat Company,
which is under contract to the U.S. Navy, assembles troop deployment boats. As part of its
research program, it completes the assembly of the first of a new model (PT109) of deployment
boats. The Navy is impressed with the PT109. It requests that Pacific Boat submit a proposal on
the cost of producing another six PT109s.
10-46
Pacific Boat reports the following cost information for the first PT109 assembled and uses a
90% cumulative average-time learning model as a basis for forecasting direct manufacturing
labor-hours for the next six PT109s. (A 90% learning curve means b = –0.152004.)
Required:
1. Calculate predicted total costs of producing the six PT109s for the Navy. (Pacific Boat will
keep the first deployment boat assembled, costed at $1,477,600, as a demonstration model
for potential customers.)
2. What is the dollar amount of the difference between (a) the predicted total costs for
producing the six PT109s in requirement 1 and (b) the predicted total costs for producing the
six PT109s, assuming that there is no learning curve for direct manufacturing labor? That is,
for (b) assume a linear function for units produced and direct manufacturing labor-hours.
SOLUTION
(30–40 min.) Cost estimation, cumulative average-time learning curve.
1. Cost to produce the 2nd through the 7th troop deployment boats:
Direct materials, 6 × $199,000 $1,194,000

Direct manufacturing labor (DML), 61,8521 × $42 2,597,784
Variable manufacturing overhead, 61,852 × $26 1,608,152
Other manufacturing overhead, 20% of DML costs 519,557
Total costs $5,919,493
1
The direct manufacturing labor-hours to produce the second to seventh boats can be calculated in several
ways, given the assumption of a cumulative average-time learning curve of 90%:
10-47
Use of table format:
90% Learning Curve Cumulative

Cumulative Cumulative Total Time:
Number of Units (X) Average Time per Unit (y): Labor Hours Labor-Hours
(1) (2) (3) = (1) × (2)
1 14,700 14,700
2 13,230 = (14,700 × 0.90) 26,460
3 12,439 37,317
4 11,907 = (13,230 × 0.90) 47,628
5 11,509 57,545
6 11,195 67,170
7 10,936 76,552
The direct labor-hours required to produce the second through the seventh boats is 76,552 – 14,700 =
61,852 hours.
Use of formula: y = aXb
where a = 14,700, X = 7, and b = – 0.152004

y = 14,700 × 7– 0.152004 = 10,936 hours
The total direct labor-hours for 7 units is 10,936 × 7 = 76,552 hours
Note: Some students will debate the exclusion of the $279,000 tooling cost. The question
specifies that the tooling “cost was assigned to the first boat.” Although Pacific Boat may well
seek to ensure its total revenue covers the $1,477,600 cost of the first boat, the concern in this
question is only with the cost of producing six more PT109s.
2. Cost to produce the 2nd through the 7th boats assuming linear function for direct labor-
hours and units produced:
Direct manufacturing labor (DML), 6 × 14,700 hrs. × $42 3,704,400
Variable manufacturing overhead, 6 × 14,700 hrs. × $26 2,293,200
The difference in predicted costs is:
Predicted cost in requirement 2
(based on linear cost function) $7,932,480
Predicted cost in requirement 1
(based on 90% learning curve) 5,919,493
Difference in favor of learning curve cost function $2,012,987
Note that the linear cost function assumption leads to a total cost that is almost 35% higher than
the cost predicted by the learning curve model. Learning curve effects are most prevalent in large
manufacturing industries such as airplanes and boats where costs can run into the millions or
hundreds of millions of dollars, resulting in very large and monetarily significant differences
10-48
between the two models. In the case of Pacific Boat, if it is in fact easier to produce additional
boats as the firm gains experience, the learning curve model is the right one to use. The firm can
better forecast its future costs and use that information to submit an appropriate cost bid to the
Navy, as well as refine its pricing plans for other potential customers.
10-41 Cost estimation, incremental unit-time learning model. Assume the same information
for the Pacific Boat Company as in Problem 10-40 with one exception. This exception is that
Pacific Boat uses a 90% incremental unit-time learning model as a basis for predicting direct
manufacturing labor-hours in its assembling operations. (A 90% learning curve means
b = –0.152004.)
Required:
1. Prepare a prediction of the total costs for producing the six PT109s for the Navy.
2. If you solved requirement 1 of Problem 10-40, compare your cost prediction there with the
one you made here. Why are the predictions different? How should Pacific Boat decide
which model it should use?
SOLUTION
(20–30 min.) Cost estimation, incremental unit-time learning model.
1. Cost to produce the 2nd through the 7th boats:

Direct manufacturing labor (DML), 71,2171 × $42 2,991,114
Variable manufacturing overhead, 71,217 × $26 1,851,642
1
The direct labor hours to produce the second through the seventh boats can be calculated via a table
format, given the assumption of an incremental unit-time learning curve of 90%:
90% Learning Curve

Cumulative Cumulative
Number of Individual Unit Time for Xth Total Time:
Units (X) Unit (y)*: Labor Hours Labor-Hours
(1) (2) (3)
1 14,700 14,700
2 13,230 = (14,700 × 0.90) 27,930
3 12,439 40,369
4 11,907 = (13,230 × 0.90) 52,276
5 11,510 63,786
6 11,195 74,981
7 10,936 85,917
*Calculated as y = aXb where a = 14,700, b = – 0.152004, and X = 1, 2, 3,. . .7.
10-49
The direct manufacturing labor-hours to produce the second through the seventh boat is 85,917 –
14,700 = 71,217 hours.
2. Difference in total costs to manufacture the second through the seventh boat under the
incremental unit-time learning model and the cumulative average-time learning model is
$6,634,979 (calculated in requirement 1 of this problem) – $5,919,493 (from requirement 1 of
Problem 10-41) = $715,486, i.e., the total costs are higher for the incremental unit-time model.
The incremental unit-time learning curve has a slower rate of decline in the time required
to produce successive units than does the cumulative average-time learning curve (see Problem
10-41, requirement 1). Assuming the same 90% factor is used for both curves:
Estimated Cumulative Direct Manufacturing Labor-Hours

Cumulative Cumulative Average- Incremental Unit-Time
Number of Units Time Learning Model Learning Model
1 14,700 14,700
2 26,460 27,930
4 47,628 52,276
7 76,552 85,917
The reason is that, in the incremental unit-time learning model, as the number of units
double, only the last unit produced has a cost of 90% of the initial cost. In the cumulative
average-time learning model, doubling the number of units causes the average cost of all the
units produced (not just the last unit) to be 90% of the initial cost.
Pacific Boat should examine its own internal records on past jobs and seek information from
engineers, plant managers, and workers when deciding which learning curve better describes the
behavior of direct manufacturing labor-hours on the production of the PT109 boats.
10-42 Regression; choosing among models. Apollo Hospital specializes in outpatient surgeries
for relatively minor procedures. Apollo is a nonprofit institution and places great emphasis on
controlling costs in order to provide services to the community in an efficient manner.
Apollo’s CFO, Julie Chen, has been concerned of late about the hospital’s consumption of
medical supplies. To better understand the behavior of this cost, Julie consults with Rhett Bratt,
the person responsible for Apollo’s cost system. After some discussion, Julie and Rhett conclude
that there are two potential cost drivers for the hospital’s medical supplies costs. The first driver
is the total number of procedures performed. The second is the number of patient-hours
generated by Apollo. Julie and Rhett view the latter as a potentially better cost driver because the
hospital does perform a variety of procedures, some more complex than others.
Rhett provides the following data relating to the past year to Julie.
10-50
Required:
1. Estimate the regression equation for (a) medical supplies costs and number of procedures and
(b) medical supplies costs and number of patient-hours. You should obtain the following
results:
Regression 1: Medical supplies costs = a + (b × Number of procedures)
Variable Coefficient Standard t-Value

Error
Constant $36,939.77 $56,404.86 0.65
Independent variable: No. of procedures $ 361.91 $ 152.93 2.37
2
Regression 2: Medical supplies costs = a + (b × Number of patient-hours)

Constant $3,654.86 $23,569.51 0.16
Independent variable: No. of patient-hours $ 56.76 $ 7.82 7.25
r2 = 0.84; Durbin-Watson statistic = 1.91
2. On different graphs plot the data and the regression lines for each of the following cost
functions:
a. Medical supplies costs = a + (b × Number of procedures)
b. Medical supplies costs = a + (b × Number of patient-hours)
3. Evaluate the regression models for “Number of procedures” and “Number of patient-hours”
as the cost driver according to the format of Exhibit 10-18 (page 406).
10-51
4. Based on your analysis, which cost driver should Julie Chen adopt for Apollo Hospital?
Explain your answer.
SOLUTION
(30 min.) Regression; choosing among models.
1. See Solution Exhibit 10-42A below.
(a) Regression Output for Medical Supplies Costs and Number of Procedures
SUMMARY OUTPUT
R Square 0.358983696
Adjusted R
Square 0.294882065
Observations 12
ANOVA
Significance
df SS MS F F
Regression 1 15730276644 1.57E+10 5.60 0.04
Residual 10 28088640022 2.81E+09
Total 11 43818916667
Standard Upper
Intercept 36939.77 56404.86 0.65 0.53 -88738.09 162617.63
X Variable 1 361.91 152.93 2.37 0.04 21.16 702.66
(b) Regression Output for Medical Supplies Costs and Number of Patient-Hours
SUMMARY OUTPUT
R Square 0.84032421
Adjusted R
Square 0.82435663
Observations 12
ANOVA
10-52
Significance
df SS MS F F
Regression 1 36822096457 3.68E+10 52.63 0.00
Residual 10 6996820210 7E+08
Total 11 43818916667
Standard Upper
Intercept 3654.86 23569.51 0.16 0.88 -48861.29 56171.00
X Variable 1 56.76 7.82 7.25 0.00 39.33 74.19
2. See Solution Exhibit 10-42B below.

Plots and Regression Lines for (a) Medical Supplies Costs and Number of Procedures and (b)
Medical Supplies Costs and Number of Patient-Hours
(a)
Apollo Hospitals
250,000
Medical supplies costs
200,000
y = 361.91x + 36940
150,000 R² = 0.35898
100,000
50,000
100 200 300 400 500 600
Number of procedures
10-53
(b)
Apollo Hospitals
250,000
Medical supplies costs
200,000 y = 56.759x + 3654.9

R² = 0.84032
150,000
100,000
50,000
1,000 1,500 2,000 2,500 3,000 3,500 4,000 4,500
Number of patient-hours
3.
Number of Setups Number of Setup Hours
Economic A positive relationship A positive relationship between
plausibility between medical supplies costs medical supplies costs and the number
and the number of procedures of patient-hours is also economically
is economically plausible. plausible, especially since the time
taken to serve patients is not uniform.
Patient-hours is more likely to capture
the true level of activity in the hospital
since it accounts for the mix of
procedures performed.
Goodness of fit r2 = 36% r2 = 84%

Standard error of regression = $52,999 Standard error of regression = $26,452
Reasonable goodness of fit. Excellent goodness of fit.
Significance of The t-value of 2.37 is significant at the The t-value of 7.25 is highly
Independent 0.05 level. It is not significant at the significant at the 0.05 and 0.01 levels.
Variables 0.01 level.
Specification Based on a plot of the data, the Based on a plot of the data, the
analysis of linearity assumption holds, but there is assumptions of linearity, constant
estimation some possibility that the constant variance, independence of residuals
assumptions variance assumption does not hold. (Durbin-Watson = 1.91), and
The Durbin-Watson statistic of 2.48 normality of residuals hold. However,
10-54
suggests the residuals are independent. inferences drawn from only 12
The normality of residuals assumption observations are not reliable.
appears to hold. However, inferences
drawn from only 12 observations are
not reliable.
4. The regression model using number of patient-hours should be used to estimate medical
supplies costs because the number of patient-hours is a more economically plausible cost driver
of medical supplies costs (compared to the number of procedures performed). The time taken to
prepare medical facilities and to actually deal with patient issues (surgery, post-procedure care,
etc.) is different for different procedures. The more complex the procedure, the more time is
taken with the patient to analyze and manage the problem, and the greater the supplies costs
incurred. As such, patient-hours might serve as a better driver of medical supplies costs. The
regression of number of patient-hours and medical supplies costs also has a better fit, a
substantially significant independent variable, and better satisfies the assumptions of the
estimation technique used.
10-43 Multiple regression (continuation of 10-42). After further discussion, Julie and Rhett
wonder if they should view both the number of procedures and number of patient-hours as cost
drivers in a multiple regression estimation in order to best understand Apollo’s medical supplies
costs.
Required:
1. Conduct a multiple regression to estimate the regression equation for medical supplies costs
using both number of procedures and number of patient-hours as independent variables. You
should obtain the following result:
Regression 3: Medical supplies costs = a + (b1 × No. of procedures) + (b2 × No. of patient-
hours)

Constant –$3,103.76 $30,406.54 –0.10
Independent variable 1: No. of procedures $ 38.24 $ 100.76 0.38
Independent variable 2: No. of patient-hours $ 54.37 $ 10.33 5.26
2
2. Evaluate the multiple regression output using the criteria of economic plausibility goodness
of fit, significance of independent variables, and specification of estimation assumptions.
3. What potential issues could arise in multiple regression analysis that are not present in simple
regression models? Is there evidence of such difficulties in the multiple regression presented
in this problem? Explain.
4. Which of the regression models from Problems 10-42 and 10-43 would you recommend Julie
Chen use? Explain.
10-55
SOLUTION
(30 min.) Multiple regression (continuation of 10-42).
1. Solution Exhibit 10-43 presents the regression output for medical supplies costs using both
number of procedures and number of patient-hours as independent variables (cost drivers).

Regression Output for Multiple Regression for Medical Supplies Costs Using Both Number of
Procedures and Number of Patient-Hours as Independent Variables (Cost Drivers)
SUMMARY OUTPUT
R Square 0.84284017
Adjusted R Square 0.80791577
Observations 12
ANOVA
Significance
df SS MS F F
Regression 2 36932343254 1.85E+10 24.13 0.00
Residual 9 6886573413 7.65E+08
Total 11 43818916667
Standard Upper
Intercept -3103.76 30406.54 -0.10 0.92 -71888.13 65680.61
X Variable 1 38.24 100.76 0.38 0.71 -189.68 266.17
X Variable 2 54.37 10.33 5.26 0.00 31.00 77.73
2.
Economic A positive relationship between medical supplies costs and each of the
plausibility independent variables (number of procedures and number of patient-hours)
is economically plausible.
Goodness of fit r2 = 84%, Adjusted r2 = 81%

Standard error of regression =$27,662
Excellent goodness of fit.
Significance of The t-value of 0.38 for number of procedures is not significant at the 0.05
Independent level. The t-value of 5.26 for number of patient-hours is significant at the
Variables 0.05 and 0.01 levels.
10-56
Specification Assuming linearity, constant variance, and normality of residuals, the
analysis of Durbin-Watson statistic of 1.96 suggests the residuals are independent.
estimation However, we must be cautious when drawing inferences from only 12
assumptions observations.
3. Multicollinearity is an issue that can arise with multiple regression but not simple regression
analysis. Multicollinearity means that the independent variables are highly correlated.
The correlation feature in Excel’s Data Analysis reveals a coefficient of correlation of

0.61 between number of procedures and number of patient-hours. This is close to the threshold
of 0.70 that is usually taken as a sign of multicollinearity problems. As evidence, note the
substantial drop in the t-value for patient-hours from 7.25 to 5.26, despite a fairly small change
in the estimated coefficient (from $56.76 to $54.37).
4. The simple regression model using the number of patient-hours as the independent variable
achieves a comparable r2 to the multiple regression model. However, the multiple regression
model includes an insignificant independent variable, number of procedures. Adding this
variable does not improve Apollo Hospital’s ability to better estimate medical supplies costs and
it also introduces multicollinearity issues. Julie should use the simple regression model with
number of patient-hours as the independent variable to estimate medical supplies costs.
10-44 Cost estimation. Hankuk Electronics started production on a sophisticated new

smartphone running the Android operating system in January 2017. Given the razor-thin margins
in the consumer electronics industry, Hankuk’s success depends heavily on being able to produce
the phone as economically as possible.
At the end of the first year of production, Hankuk’s controller, Inbee Kim, gathered data on
its monthly levels of output, as well as monthly consumption of direct labor-hours (DLH). Inbee
views labor-hours as the key driver of Hankuk’s direct and overhead costs. The information
collected by Inbee is provided below:
10-57
Required:
1. Inbee is keen to examine the relationship between direct labor consumption and output
levels. She decides to estimate this relationship using a simple linear regression based on the
monthly data. Verify that the following is the result obtained by Inbee:
Regression 1: Direct labor-hours = a + (b × Output units)

Constant 345.24 589.07 0.59
Independent variable: Output units 0.71 0.93 0.76
3. Inbee estimates that Hankuk has a variable cost of $17.50 per direct labor-hour. She expects
that Hankuk will produce 650 units in the next month, January 2018. What should she budget
as the expected variable cost? How confident is she of her estimate?
SOLUTION
(30 min.) Cost estimation.
1. Here is the summary output for the monthly regression of Direct Labor Hours on Output
Units for Hankuk Electronics:
SUMMARY OUTPUT
R Square 0.054457
Adjusted R
Square -0.0400973
Observations 12
ANOVA
Significance
df SS MS F F
Regression 1 24483.86 24483.86 0.575933 0.465422344
Residual 10 425116.1 42511.61
Total 11 449600
Standard Upper
Intercept 345.24 589.07 0.59 0.57 -967.29 1657.77
X Variable 1 0.71 0.93 0.76 0.47 -1.37 2.79
10-58
2. The plot and regression line for monthly direct labor hours on monthly output for Hankuk
Electronics are given below:
Hankuk Electronics
1,600
1,400 y = 0.7091x + 345.24

Direct Labor Hours
R² = 0.05446
1,200
1,000
800
600
400
450 500 550 600 650 700 750
Output (Units)
Economic A positive relationship between direct labor hours and monthly output is
plausibility economically plausible since increased levels of production should lead to
the consumption of greater amounts of direct labor.
Goodness of fit r2 = 5.45%, Adjusted r2 = - 4%

Terrible fit; in fact, there is no evidence of a linear relationship between the
dependent and independent variables. At least one data point represents a
significant outlier.
Significance of The t-value of 0.76 for output units is not significant at the 0.05 level.
Independent
Variables
3. Given Inbee’s expectation that Hankuk will produce 650 units in January 2018, her best
estimate given the linear regression above is that Hankuk will use:
345.24 + (0.71 × 650 units) = 806.74 direct labor hours.
At an estimated variable cost of $17.50 per direct labor-hour, this implies that Inbee should
budget
806.74 × $17.50 = $14,118
for direct labor costs for January 2018.
10-59
Note that 650 units is in the range of output values that were used to find the regression equation,
and therefore falls in the range of predictability for this model. However, there is substantial
uncertainty around the cost estimate of $14,118. In particular, this predicted value relies on the
regression point estimate of 0.71 for the marginal impact of output on labor hours. But, the 95%
confidence interval for the slope of the regression ranges all the way from -1.37 to 2.79, and the
predicted cost would vary accordingly. One cannot reject the null hypothesis that output levels
have no impact on labor consumption, leaving the budgeted cost estimate a highly speculative
one!
10-45 Cost estimation, learning curves (continuation of 10-44). Inbee is concerned that she
still does not understand the relationship between output and labor consumption. She consults
with Jim Park, the head of engineering, and shares the results of her regression estimation. Jim
indicates that the production of new smartphone models exhibits significant learning effects—as
Hankuk gains experience with production, it can produce additional units using less time. He
suggests that it is more appropriate to specify the following relationship:
y = axb
where x is cumulative production in units, y is the cumulative average direct labor-hours per unit
(i.e., cumulative DLH divided by cumulative production), and a and b are parameters of the
learning effect.
To estimate this, Inbee and Jim use the original data to calculate the cumulative output and
cumulative average labor-hours per unit for each month. They then take natural logarithms of
these variables in order to be able to estimate a regression equation. Here is the transformed data:
10-60
Required:
1. Estimate the relationship between the cumulative average direct labor-hours per unit and
cumulative output (both in logarithms). Verify that the following is the result obtained by
Inbee and Jim:
Regression 1: Ln (Cumulative avg DLH per unit) = a + [b × Ln (Cumulative Output)]

Constant 2.087 0.024 85.44
Independent variable: Ln (Cum Output) –0.208 0.003 –69.046
2
3. Verify that the estimated slope coefficient corresponds to an 86.6% cumulative average-time
learning curve.
4. Based on this new estimation, how will Inbee revise her budget for Hankuk’s variable cost
for the expected output of 650 units in January 2018? How confident is she of this new cost
estimate?
SOLUTION
(30 min.) Cost estimation, learning curves (continuation of 10-44).
1. Here is the summary output for the monthly regression of the natural log of Cumulative
Average Direct Labor-Hours per Unit on the natural logarithm of Cumulative Output:
SUMMARY OUTPUT
R Square 0.9979068
Adjusted R
Square 0.9976975
Observations 12
ANOVA
Significance
df SS MS F F
9.89803E-
Regression 1 0.263368 0.263368 4767.34 15
Residual 10 0.000552 5.52E-05
Total 11 0.26392
10-61
Standard Upper
Intercept 2.09 0.02 85.44 0.00 2.03 2.14
X Variable 1 -0.21 0.00 -69.05 0.00 -0.21 -0.20
2. The plot of the data and the regression line estimated above are provided next.
Hankuk Electronics
0.800 y = -0.2079x + 2.0876
Log of Cumulative Average DLH per unit
R² = 0.99791
0.700
0.600
0.500
0.400
0.300
0.200
6.000 6.500 7.000 7.500 8.000 8.500 9.000
Log of Cumulative Output
Economic A negative relationship between cumulative average direct-labor hours per

plausibility unit and cumulative output (in natural logarithms) is economically
plausible and reflects the presence of learning effects. Specifically, as the
firm gains experience via production, it becomes more efficient and is able
to use fewer direct labor hours to make each unit of product.
Goodness of fit r2 = 99.8%, Adjusted r2 = 99.8%

Unparalleled goodness of fit. Virtually perfect linear fit in logarithms.
Significance of The t-value of -69.05 for the logarithm of cumulative output is significant
Independent at all conventional levels. The t-value for the intercept (85.44) is highly
Variables significant as well.
3. The original learning curve specification, y = axb is mathematically identical to the following
log-linear specification:
Ln y = Ln a + b × Ln x
The regression equation we have estimated,
Ln (Cumulative avg DLH per unit) = a + (b × Ln (Cumulative Output))
10-62
is precisely the above specification, and in particular the slope coefficient directly yields the “b”
from the learning curve equation. We know therefore that for Hankuk electronics, b = -0.208. As
explained in Exhibit 10-10, this value is related to the learning curve percentage as follows:
b = Ln (learning-curve % in decimal form)/Ln 2, or

-0.208 = Ln (learning-curve % in decimal form)/0.693, or
Ln (learning-curve % in decimal form) = -0.208 × 0.693 = -0.144.
As the exponent of -0.144 is 0.8659, this implies that Hankuk is experiencing an 86.6%
cumulative average-time learning curve.
4. With an additional 650 units in January 2018, Hankuk’s cumulative output will go from 7,527
at the end of December 2016 to 8,177 (7,527 + 650). As Ln (8,177) = 9.0091, the cumulative
average direct-labor hours in logarithmic terms are given by:
2.0876 – 0.2079 × 9.0091 = 0.2146.
The cumulative direct-labor hours per unit therefore equals Exp (0.2146) = 1.2394. This implies
a total direct labor hours of 1.2394 × 8,177 = 10,134 by the end of January. As Hankuk has
used a total of 9,480 direct labor hours at the end of December 2017, the incremental hours
needed in January therefore are 654 (10,134 – 9,480). At $17.50 per labor hour, this suggest that
Inbee should budget
654 × $17.50 = $11,445
for direct labor costs for January 2018.
While 9.0091 is outside the range of cumulative output values (measured in logarithms) used to
find the regression equation, unless there has been a structural break in the experience curve
Hankuk is facing, it is highly likely that its January costs will be in the neighborhood of $11,445.
The reason is that the estimated regression line is close to perfect and has a standard error close
to zero. There is virtually no uncertainty around the coefficient estimates. The slope coefficient,
for example, has a point estimate of -0.2079, and a narrow 95% confidence interval between -
0.2146 and -0.2012. Using either of those estimates would make barely any difference to the
predicted cost for the month of January 2018.
10-46 Interpreting regression results, matching time periods. Nandita Summers works at
Modus, a store that caters to fashion for young adults. Nandita is responsible for the store’s
online advertising and promotion budget. For the past year, she has studied search engine
optimization and has been purchasing keywords and display advertising on Google, Facebook,
and Twitter. In order to analyze the effectiveness of her efforts and to decide whether to continue
online advertising or move her advertising dollars back to traditional print media, Nandita
collects the following data:
10-63
Required:
1. Nandita performs a regression analysis, comparing each month’s online advertising expense
with that month’s revenue. Verify that she obtains the following result:
Revenue = $51,999.64 – (0.98 × Online advertising expense)

Constant $51,999.64 7,988.68 6.51
Independent variable: Online advertising expense –0.98 1.99 –0.49
2
2. Plot the preceding data on a graph and draw the regression line. What does the cost formula
indicate about the relationship between monthly online advertising expense and monthly
revenues? Is the relationship economically plausible?
3. After further thought, Nandita realizes there may have been a flaw in her approach. In
particular, there may be a lag between the time customers click through to the Modus website
and peruse its social media content (which is when the online ad expense is incurred) and the
time they actually shop in the physical store. Nandita modifies her analysis by comparing
each month’s sales revenue to the advertising expense in the prior month. After discarding
September revenue and August advertising expense, show that the modified regression yields
the following:
Revenue = $28,361.37 + (5.38 × Online advertising expense)
10-64
Constant $28,361.37 5,428.69 5.22
Independent variable: Previous month’s online 5.38 1.31 4.12
advertising expense r2 = 0.65; Durbin-Watson
statistic = 1.71
4. What does the revised formula indicate? Plot the revised data on a graph. Is this relationship
economically plausible?
5. Can Nandita conclude that there is a cause-and-effect relationship between online advertising
expense and sales revenue? Why or why not?
SOLUTION
(25 min.) Interpreting regression results, matching time periods
1. Here is the summary output for the monthly regression of Sales Revenue on Online
Advertising Expense:
SUMMARY OUTPUT
Multiple R 0.15
R Square 0.02
Adjusted R
Square -0.07
Observations 12.00
ANOVA
Significance
df SS MS F F
Regression 1 33972689.79 33972690 0.242451 0.633072
Residual 10 1401216525 1.4E+08
Total 11 1435189215
Standard Upper
Intercept 51999.64 7988.68 6.51 0.00 34199.74 69799.54
X Variable 1 -0.98 1.99 -0.49 0.63 -5.41 3.45
2. SOLUTION EXHIBIT 10-46A presents the data plot for the initial analysis. The formula
of Sales Revenue = $52,000 – (0.98 × Online advertising expense) indicates that there is a fixed
amount of revenue each month of $52,000, which is reduced by 0.98 times that month’s online
advertising expense. This relationship is not economically plausible, as advertising would not
10-65
reduce revenue. The data points do not appear linear, and the r-square of 0.02 indicates a very
weak goodness of fit (in fact, almost no fit at all).
SOLUTION EXHIBIT 10-46 A

Plot and Regression Line for Sales Revenue and Online Advertising Expense
$70,000
$65,000
$60,000
$55,000
Sales Revenue
$50,000
$45,000
y = -0.9789x + 52000
$40,000 R² = 0.02367
$35,000
$30,000
$25,000
$20,000
$0 $1,000 $2,000 $3,000 $4,000 $5,000 $6,000 $7,000
Online Advertising Expense
3. Here is the summary output for the regression of monthly Sales Revenue on the prior
month’s Online Advertising Expense:
SUMMARY OUTPUT
Multiple R 0.808588
R Square 0.653815
Adjusted R
Square 0.61535
Observations 11
ANOVA
Significance
df SS MS F F
Regression 1 9.29E+08 929262059 16.99763 0.002587
Residual 9 4.92E+08 54670085
Total 10 1.42E+09
Standard Upper
Intercept 28361.37 5428.687 5.2243522 0.000546 16080.83 40641.91
X Variable 1 5.381665 1.305336 4.1228186 0.002587 2.428789 8.33454
10-66
3. SOLUTION EXHIBIT 10-46 B presents the data plot for the revised analysis. The
formula of Sales Revenue = $28,361 + (5.38 × Online Advertising Expense) indicates that there
is a fixed amount of revenue each month of $28,361, which increases by 5.38 times the prior
month’s advertising expense in the online channel. This relationship is economically plausible.
One would expect a positive correlation between advertising expense and (future) sales revenue.
The slope coefficient of 5.38 has a t stat of 4.12 indicating that it is statistically significant at the
5% level. In the revised analysis, there is improved linearity in the data points, and the r-square
of 0.65 indicates a much stronger goodness of fit.

Plot and Regression Line for Sales Revenue and Previous Month Online Advertising
70,000
65,000
60,000
Sales Revenue
55,000 y = 5.3817x + 28361

R² = 0.65381
50,000
45,000
40,000
35,000
30,000
0 1,000 2,000 3,000 4,000 5,000 6,000 7,000
Online Advertising Expense (Prior Month)
4. Nandita must be very careful about making conclusions regarding cause and effect. Even
a strong goodness of fit does not prove a cause and effect relationship. The independent and
dependent variables could both be caused by a third factor, or the correlation could be simply
coincidental. However, there is enough of a correlation in the revised analysis for Nandita to
make a meaningful presentation to the store’s owner.
10-47 Purchasing department cost drivers, activity-based costing, simple regression

analysis. Perfect Fit operates a chain of 10 retail department stores. Each department store makes
its own purchasing decisions. Carl Hart, assistant to the president of Perfect Fit, is interested in
better understanding the drivers of purchasing department costs. For many years, Perfect Fit has
allocated purchasing department costs to products on the basis of the dollar value of merchandise
purchased. A $100 item is allocated 10 times as many overhead costs associated with the
purchasing department as a $10 item.
Hart recently attended a seminar titled “Cost Drivers in the Retail Industry.” In a presentation
at the seminar, Kaliko Fabrics, a leading competitor that has implemented activity-based costing,
reported number of purchase orders and number of suppliers to be the two most important cost
drivers of purchasing department costs. The dollar value of merchandise purchased in each
purchase order was not found to be a significant cost driver. Hart interviewed several members
10-67
of the purchasing department at the Perfect Fit store in Miami. They believed that Kaliko
Fabrics’ conclusions also applied to their purchasing department.
Hart collects the following data for the most recent year for Perfect Fit’s 10 retail department
stores:
Hart decides to use simple regression analysis to examine whether one or more of three variables
(the last three columns in the table) are cost drivers of purchasing department costs. Summary
results for these regressions are as follows:
Regression 1: PDC = a + (b × MP$)

Constant $1,041,421 $346,709 3.00
Independent variable 1: MP$ 0.0031 0.0038 0.83
Regression 2: PDC = a + (b × No. of POs)

Constant $722,538 $265,835 2.72
Independent variable 1: No. of POs $ 159.48 $ 64.84 2.46
2
Regression 3: PDC = a + (b × No. of Ss)
10-68
Constant $828,814 $246,571 3.36
Independent variable 1: No. of Ss $ 3,816 $ 1,698 2.25
2
1. Compare and evaluate the three simple regression models estimated by Hart. Graph each one.
Also, use the format employed in Exhibit 10-18 (page 406) to evaluate the information.
2. Do the regression results support the Kaliko Fabrics’ presentation about the purchasing
department’s cost drivers? Which of these cost drivers would you recommend in designing
an ABC system?
3. How might Hart gain additional evidence on drivers of purchasing department costs at each
of Perfect Fit’s stores?
SOLUTION
(40–50 min.) Purchasing Department cost drivers, activity-based costing, simple

regression analysis.
The problem reports the exact t-values from the computer runs of the data. Because the
coefficients and standard errors given in the problem are rounded to three decimal places,
dividing the coefficient by the standard error may yield slightly different t-values.
1. Plots of the data used in Regressions 1 to 3 are in Solution Exhibit 10-47A. See Solution
Exhibit 10-47B for a comparison of the three regression models.
2. Both Regressions 2 and 3 are well-specified regression models. The slope coefficients on
their respective independent variables are significantly different from zero. These results support
the Kaliko Fabrics’ presentation in which the number of purchase orders and the number of
suppliers were reported to be drivers of purchasing department costs.
In designing an activity-based cost system, Perfect Fit should use number of purchase
orders and number of suppliers as cost drivers of purchasing department costs. As the chapter
appendix describes, Perfect Fit can either (a) estimate a multiple regression equation for
purchasing department costs with number of purchase orders and number of suppliers as cost
drivers, or (b) divide purchasing department costs into two separate cost pools, one for costs
related to purchase orders and another for costs related to suppliers, and estimate a separate
relationship for each cost pool.
3. Guidelines presented in the chapter could be used to gain additional evidence on cost
drivers of purchasing department costs.
1. Use physical relationships or engineering relationships to establish cause-and-effect

links. Hart could observe the purchasing department operations to gain insight into
how costs are driven.
10-69
2. Use knowledge of operations. Hart could interview operating personnel in the purchasing
department to obtain their insight on cost drivers.

Regression Lines of Various Cost Drivers for Purchasing Dept. Costs for Perfect Fit
$2,500,000
y = 0.0031x + 1E+06
Purchasing Department Costs
$2,000,000 R² = 0.07981
$1,500,000
$1,000,000
$500,000
$0
0 50,000,000 100,000,000 150,000,000
Dollar Value of Merchandise Purchased
$2,500,000
y = 159.48x + 722538
R² = 0.43064
$2,000,000
$1,500,000
$1,000,000
$500,000
$0
0 1,000 2,000 3,000 4,000 5,000 6,000 7,000 8,000
Number of Purchase Orders
10-70
$2,500,000

$2,000,000 y = 3815.7x + 828814
R² = 0.38685
$1,500,000
$1,000,000
$500,000
$0
0 50 100 150 200 250
Number of Suppliers

Comparison of Alternative Cost Functions for Purchasing Department
Costs Estimated with Simple Regression for Perfect Fit
Regression 1 Regression 2 Regression 3

Criterion PDC = a + (b × MP$) PDC = a + (b × # of POs) PDC = a + (b × # of Ss)
1. Economic Result presented at Economically plausible. Economically plausible.
Plausibility seminar by Kaliko The higher the number of Increasing the number of
Fabrics found little purchase orders, the more suppliers increases the
support for MP$ as a tasks undertaken. costs of certifying
driver. Purchasing vendors and managing
personnel at the the Perfect Fit-supplier
Miami store believe relationship.
MP$ is not a
significant cost driver.
2. Goodness of fit r2 = 0.08. Poor r2 = 0.43. Reasonable r2 = 0.39. Reasonable

goodness of fit. goodness of fit. goodness of fit.
3. Significance of t-value on MP$ of t-value on # of POs of 2.46 t-value on # of Ss of 2.25

Independent 0.83 is insignificant. is significant. is significant.
Variables
10-71
4. Specification
Analysis
A. Linearity Appears questionable Appears reasonable. Appears reasonable.

within the but no strong evidence
relevant range against linearity.
B. Constant Appears questionable, Appears reasonable. Appears reasonable.

variance of but no strong evidence
residuals against constant
variance.
C. Independence Durbin-Watson Durbin-Watson Durbin-Watson

of residuals Statistic = 2.41 Statistic = 1.97 Statistic = 2.01
Assumption of Assumption of Assumption of
independence is not independence is not independence is not
rejected. rejected. rejected.
D. Normality of Data base too small to Data base too small to Data base too small to
residuals make reliable make reliable inferences. make reliable inferences.
inferences.
10-48 Purchasing department cost drivers, multiple regression analysis (continuation of

10-47). Carl Hart decides that the simple regression analysis used in Problem 10-47 could be
extended to a multiple regression analysis. He finds the following results for two multiple
regression analyses:
Regression 4: PDC = a + (b1 × No. of POs) + (b2 × No. of Ss)

Constant $484,522 $256,684 1.89
2
Regression 5: PDC = a + (b1 × No. of POs) + (b2 × No. of Ss) + (b3 × MP$)

Constant $483,560 $312,554 1.55
10-72
Independent variable 3: MP$ 0.00002 0.0029 0.01
2
The coefficients of correlation between combinations of pairs of the variables are as follows:
PDC MP$ No. of POs

MP$ 0.28
No. of POs 0.66 0.27
No. of Ss 0.62 0.30 0.29
Required
1. Evaluate regression 4 using the criteria of economic plausibility, goodness of fit,
significance of independent variables, and specification analysis. Compare regression 4
with regressions 2 and 3 in Problem 10-47. Which one of these models would you
recommend that Hart use? Why?
2. Compare regression 5 with regression 4. Which one of these models would you recommend
that Hart use? Why?
3. Hart estimates the following data for the Baltimore store for next year: dollar value of
merchandise purchased, $78,500,000; number of purchase orders, 4,100; number of
suppliers, 110. How much should Hart budget for purchasing department costs for the
Baltimore store for next year?
4. What difficulties do not arise in simple regression analysis that may arise in multiple
regression analysis? Is there evidence of such difficulties in either of the multiple regressions
presented in this problem? Explain.
5. Give two examples of decisions in which the regression results reported here (and in Problem
10-47) could be informative.
SOLUTION
(30–40 min.) Purchasing Department cost drivers, multiple regression analysis

(continuation of 10-47).
The problem reports the exact t-values from the computer runs of the data. Because the
coefficients and standard errors given in the problem are rounded to three decimal places,
dividing the coefficient by the standard error may yield slightly different t-values.
1. Regression 4 is a well-specified regression model:
Economic plausibility: Both independent variables are plausible and are supported by the
findings of the Kaliko Fabrics study.
Goodness of fit: The r2 of 0.64 indicates an excellent goodness of fit.
10-73
Significance of independent variables: The t-value on # of POs is 2.19 while the t-value on # of
Ss is 1.99. These t-values are either significant or border on significance.
Specification analysis: Results are available to examine the independence of residuals

assumption. The Durbin-Watson statistic of 1.91 indicates that the assumption of independence
is not rejected.
Regression 4 is consistent with the findings in Problem 10-47 that both the number of
purchase orders and the number of suppliers are drivers of purchasing department costs.
Regressions 2, 3, and 4 all satisfy the four criteria outlined in the text. Regression 4 has the best
goodness of fit (0.64 for Regression 4 compared to 0.43 and 0.39 for Regressions 2 and 3,
respectively). Most importantly, it is economically plausible that both the number of purchase
orders and the number of suppliers drive purchasing department costs. We would recommend
that Hart use Regression 4 over Regressions 2 and 3.
2. Regression 5 adds an additional independent variable (MP$) to the two independent

variables in Regression 4. This additional variable (MP$) has a t-value of -0.01, implying its
slope coefficient is insignificantly different from zero. The r2 in Regression 5 (0.64) is the same
as that in Regression 4 (0.64), implying the addition of this third independent variable adds close
to zero explanatory power. In summary, Regression 5 adds very little to Regression 4. We would
recommend that Hart use Regression 4 over Regression 5.
3. Budgeted purchasing department costs for the Baltimore store next year are
$484,522 + ($126.66 × 4,100) + ($2,903 × 110) = $1,323,158
4. Multicollinearity is a frequently encountered problem in cost accounting; it does not arise

in simple regression because there is only one independent variable in a simple regression. One
consequence of multicollinearity is an increase in the standard errors of the coefficients of the
individual variables. This frequently shows up in reduced t-values for the independent variables
in the multiple regression relative to their t-values in the simple regression:
t-value from
t-value in Simple Regressions
Variables Multiple Regression in Problem 10-47
Regression 4:
# of POs 2.19 2.46
# of Ss 1.99 2.25
Regression 5:
# of POs 1.99 2.46
# of Ss 1.79 2.25
MP$ -0.01 0.83
10-74
The decline in the t-values in the multiple regressions is consistent with some (but not very high)
collinearity among the independent variables. Pairwise correlations between the independent
variables are:
Correlation
# of POs and # of Ss 0.29
# of POs and MP$ 0.27
# of Ss and MP$ 0.30
There is no evidence of difficulties due to multicollinearity in Regressions 4 and 5.
5. Decisions in which the regression results in Problems 10-47 and 10-48 could be useful
are
Cost management decisions: Perfect Fit could restructure relationships with the suppliers so that
fewer separate purchase orders are made. Alternatively, it may aggressively reduce the number
of existing suppliers.
Purchasing policy decisions: Perfect Fit could set up an internal charge system for individual
retail departments within each store. Separate charges to each department could be made for each
purchase order and each new supplier added to the existing ones. These internal charges would
signal to each department ways in which their own decisions affect the total costs of Perfect Fit.
Accounting system design decisions: Perfect Fit may want to discontinue allocating purchasing
department costs on the basis of the dollar value of merchandise purchased. Allocation bases
that better capture cause-and-effect relations at Perfect Fit are the number of purchase orders and
the number of suppliers.
Try It! 10-1
a. y = $1.70X
b. y = $8,000
c. y = $80 + $2.00X
d. y = $1,000 + $12X
Try It! 10-2
The highest and lowest observations of the cost driver correspond to 5,850 hours and 3,000
hours, respectively. Using those data points:
a. Slope = ($69,850 − $38,500) / (5,850 − 3,000)

= $31,350/2,850
= $11
10-75
b. $69,850 = Constant + ($11 × 5,850)
Constant = $5,500
c. y = $5,500 + ($11 × Hours)
d. y = $5,500 + ($11 × 3,100) = $39,600
Try It! 10-3
a. Unit Hours Cumulative Cumulative Average

1 6 6 6
2 4.8 10.8 5.4
Learning percentage = 5.4/6.0 = 0.90

b
b. y =a×X
− 0.1520
=6×8
= 4.37 hours
(or) Cumulative average time for:

1 unit = 6 hours
2 units = 6 × 0.9 = 5.4 hours
4 units = 5.4 × 0.9 = 4.86 hours
8 units = 4.86 × 0.9 = 4.37 hours
Therefore, total time to build 8 units = 8 × 4.37 = 34.96 hours.
c. Total time = 34.96 hours

Manufacturing overhead charge = 34.96 × $25 = $624
d. As production doubles from 1 to 2 units, the incremental time of the second unit relative to
the first is 4.8 hours/6 hours = 0.8.
Therefore, this represents an 80% learning curve under the incremental unit-time learning model.
b
e. y = a × X
− 0.3219
= 6 × 16
= 2.458 hours
(or) Time for:

1st unit = 6 hours
2nd unit = 6 × 0.8 = 4.8 hours
4th unit = 4.8 × 0.8 = 3.84 hours
8th unit = 3.84 × 0.8 = 3.072 hours
16th unit = 3.072 × .8 = 2.458 hours
10-76
Try It! 10-4
a, b and c.
Plot and Regression Line of Sales Revenues on Promotional Costs
850,000
800,000
y = 14.184x + 397163
750,000
R² = 0.6271
Sales Revenues
700,000
650,000
600,000
550,000
500,000
450,000
400,000
5,000 10,000 15,000 20,000 25,000
Promotional Costs
The above plot includes the regression line of sales revenues on promotional costs. Here are the
details, from carrying out a regression analysis in Excel:
Sales Revenues = $397,163 + (14.18 × Promotional Costs)

Constant $397,163.12 59,169.06 6.71
Independent variable: 14.18 3.87 3.67
Promotional costs
We evaluate the estimated regression equation using the criteria of economic plausibility,
goodness of fit, and slope of the regression line.
Economic plausibility. Promotional costs would appear to be a plausible cost driver of sales
revenues. Restaurants frequently use promotional activities such as advertising, sponsoring of
local events, etc. to engender interest among potential clients and increase their patronage.
Goodness of fit. As seen in the plot, the regression line fits the data well. The vertical differences
between actual and predicted revenues are reasonably small. This indicates that promotional
10-77
costs are related to restaurant revenues. An r-squared value of 0.627 indicates that almost 63%
of the change in revenues can be explained by the change in promotional costs.
Slope of regression line. The slope of the regression line is relatively steep. Given the small
scatter of the observations around the line, this indicates that, on average, restaurant revenues
increase with newspaper advertising at a slope of 14.18. The t-value of 3.67 is statistically
significant at the 0.05 and 0.01 levels.
d. The increase in sales revenues for each $1,000 spent on promotion, within the relevant
range, is:
$14.184 × $1,000 = $14,184.
10-78

HORN Ch10 SM

Uploaded by

Copyright:

Available Formats

HORN Ch10 SM

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

HORN Ch10 SM

Uploaded by

Copyright:

Available Formats

CHAPTER 10

DETERMINING HOW COSTS BEHAVE

The two assumptions are

10-2 Describe three alternative linear cost functions.

Three alternative linear cost functions are

10-5 Name four approaches to estimating a cost function.

Four approaches to estimating a cost function are

The six steps are

Four key assumptions examined in specification analysis are

10-18 In regression analysis, the coefficient of determination:

Choice "c" is incorrect. This is not a use of the coefficient of determination.

a. The coefficient of determination is negative.

Month Miles Driven Delivery Costs

(10 min.) Estimating a cost function.

Difference in delivery costs

Constant = Total cost – (Slope coefficient × Quantity of cost driver)

= $10,000 – ($0.60 × 12,000) = $2,800

= $13,000 – ($0.60 × 17,000) = $2,800

The cost function based on the two observations is

Delivery costs = $2,800 + $0.60 × Miles driven

¡ Contract 1: $45 for the day

(15 min.) Identifying variable-, fixed-, and mixed-cost functions.

1. See Solution Exhibit 10-22.

3. Contract Cost Function

SOLUTION EXHIBIT 10-22

Contract 2: Mixed Costs

10-23 Various cost-behavior patterns. (CPA, adapted).

1. Annual depreciation of equipment, where the amount of depreciation charged is computed by

First 1,000,000 gallons or less $1,000 flat fee

(20 min.) Various cost-behavior patterns.

a. Direct material costs

Maintenance Machine Health Number of Shipping Units

(20 min.) Account analysis, high-low

$5,271 = a fixed cost + ($2.10 × number of machine hours used)

Therefore, Stein’s cost formula for monthly maintenance cost is:

3. The shipping rate is $3.60 per unit shipped

Adding them together we get:

Fixed operating costs = $4,200 (maintenance) + $8,600 (health insurance) = $12,800

4. Estimated operating cost for November:

Account Classification Amount

(30 min.) Account analysis method.

1. Manufacturing cost classification for 2017:

Total manufacturing cost for 2017 = $948,750

Variable costs in 2018:

Direct materials $4.00 5% $0.20 $4.20 $336,000

Fixed and total costs in 2018:

Direct materials $ 0 0% $ 0 $ 0 $336,000 $ 336,000

(15–20 min.) Estimating a cost function, high-low method.

Cost Driver: Operating Annual

2. The constant a (estimated as $100,000) represents the fixed costs of operating a

Week Customer-Service Department Costs Number of Service Reports

(20 min.) Estimating a cost function, high-low method.

Customer-service department costs = a + b (number of service reports)

3. Other possible cost drivers of customer-service department costs are:

Customer-Service Department Costs

Total Overhead Costs Physician Contact Hours Billed to Patients

Difference in overhead costs

= ($125,000 - $105,000)/(300 – 200)