chapter

3
Analysing and predicting
costs
Chapter learning objectives

After completing this chapter, you should be able to:

• calculate appropriate costs having identified cost behaviour.

75
Analysing and predicting costs

1 Session content diagram

2 Cost behaviour

In the previous chapter we saw that understanding costs and cost behaviour
is important. Managers must be able to identify if costs are fixed, variable,
stepped or semi­variable. If this cost behaviour is understood then
managers are able to estimate or predict costs going forward.

We will use the following example to show how managers can identify the
cost behaviour.

Illustration 1 – Identifying cost behaviour

Consider the following costs for a manufacturing company:

100 units 200 units

Material $500 $1,000
Labour $1,000 $2,000
Rent $2,000 $2,000
Electricity $700 $900

To determine the type of cost, consider the cost behaviour over a range
of activity levels.

Material: Material would normally be a variable cost – the total cost is

increasing as the number of units increases. To check if it is a linear
variable cost, divide the total cost by the number of units and the unit
cost should be the same for each level:

Material is $500 for 100 units, therefore $5 per unit, and $1,000 for 200
units, therefore $5 per unit. This suggests that material is a variable
cost.

76
 chapter 3

Labour: If you look at labour, you can see that labour is a variable
cost of $10 per unit. ($1,000 ÷ 100 = $10 and $2,000 ÷ 200 = $10)

Rent: Rent is normally a fixed cost and in the example the total rent cost
is $2,000 for each level of activity. This suggests that rent is a fixed
cost.

Electricity: For electricity the total cost is increasing as the number of

units increase, but if we work out the unit cost we can see that it varies at
each level. Electricity is $700 for 100 units, therefore $7 per unit and
$900 for 200 units, therefore $4.50 per unit. This suggests that
electricity is a semi­variable cost.

Test your understanding 1

The following data have been collected for four cost types, W, X, Y and
Z, at two activity levels:

Cost Cost
100 units 140 units
Cost type $ $
W 8,000 10,560
X 5,000 5,000
Y 6,500 9,100
Z 6,700 8,580

Where V = variable, SV = semi­variable and F = fixed, assuming

linearity, the four cost types W, X, Y and Z are, respectively:

W X Y Z
A V F SV V
B SV F V SV
C V F V V
D SV F SV SV

Analysing semi­variable costs

A semi­variable cost is made up of a fixed and a variable element.

Semi­variable costs, such as electricity, contain both a fixed and a variable

element. We know that the fixed element will remain the same and the
variable element will vary directly with the activity level but first we will have to
split the cost into these two components.

77
Analysing and predicting costs

Note: Total cost = Total fixed cost + Total variable cost

(Total variable cost = Variable cost per unit × Number of units)

Past records of costs and their associated activity levels are usually used to
carry out the analysis. Your Fundamentals of Management Accounting
syllabus requires you to know how to use three common methods for
separating the fixed and variable elements and predicting total costs:

(1) The high–low method

(2) The scatter graph (line of best fit) method
(3) Regression analysis

All three methods attempt to identify the relationship between the activity
levels and the associated cost by analysing past data. Once these
relationships are known, predictions can be made about costs going
forward.

3 The high–low method

This method picks out the highest and lowest activity levels from the
available data and investigates the change in cost which has occurred
between them. The highest and lowest points are selected to try to use the
greatest possible range of data.

From this, the variable cost per unit and the fixed cost element can be
calculated.

Illustration 2 – The high­low method

An entity has recorded the following data for a semi­variable cost:

Activity level Cost incurred

Month (units) ($)
January 1,800 36,600
February 2,450 41,150
March 2,100 38,700
April 2,000 38,000
May 1,750 36,250
June 1,950 37,650

The highest activity level occurred in February and the lowest in May.

Always select the highest and lowest activity level.

Since the amount of fixed cost incurred in each month is constant, the
extra cost resulting from the activity increase must be the variable cost.

78
 chapter 3

Activity level
(units) $
February 2,450 41,150
May 1,750 36,250
——— ———
Increase 700 4,900
——— ———

The extra variable cost for 700 units is $4,900. We can now calculate the
variable cost per unit:

Change in cost
Variable cost per unit = ——–—————
Change in units

Variable cost per unit = $4,900/700 = $7 per unit

Substituting this back in to the data for February, we can determine the
amount of fixed cost:

February $
Total cost 41,150
Variable cost (2,450 units × $7) 17,150
————
Therefore, fixed cost per month $24,000
————

Note: The calculation for fixed costs could also have used the data for
May.

We now know that the semi­variable cost has a fixed monthly amount of
$24,000 and a variable cost per unit of $7.

Advantages and disadvantages of high­low method

Advantages

• it is easy to understand and easy to use.

Disadvantages

• it relies on historical cost data and assumes this data can reliably
predict future costs.
• it assumes that the activity level is the only factor affecting costs.

79
Analysing and predicting costs

• it uses only two values (highest and lowest) to predict future costs and
these results may be distorted because of random variations which may
have occurred.
• it assumes a linear relationship between the two variables.

4 The scatter graph (line of best fit) method

With the high­low method only two pairs of data are used and a linear
relationship is assumed. However these two extreme points may hide the
true relationship between the variables.

The second method for analysing and predicting costs is the scatter graph
(line of best fit) method. This method takes account of all available historical
data. It is a graphical method which uses the equation of a straight line.

First we should consider the equation of a straight line.

When drawing a graph It is good practice to set the variables so that the x­
axis always shows the independent variable, i.e. that variable which is not
affected by the other variable. The y­axis should always represent the
dependent variable, i.e. that variable which depends on the other. A change
in the value of the independent variable will cause the dependent variable to
change.

80
 chapter 3

To use the scatter graph method:

(1) First a graph is drawn which plots all available pairs of data.

(2) The result is known as a scatter diagram, scatter graph or sometimes a

scatter plot and is a visual way of determining if there might be a (linear)
relationship between the variables x and y. If it looks as though there is
such a relationship, we can then go on to calculate the correlation
coefficient which measures the strength of the relationship between the
two variables. This will be looked at later in the chapter.
(3) Where a clear relationship can be seen, a line of best fit can be drawn
by eye. This is the line which, in the judgement of the user, appears to
be the best representation of the gradient of the sets of points on the
graph.

This is demonstrated below:

In this example, 12 pairs of data have been plotted on the graph. From
these observations, a line of best fit has been drawn to represent the
relationship between the two variables. This is shown as a solid line.
The line has then been extrapolated out towards the y axis and
intersects the y axis at $200.

The inaccuracies involved in drawing the line of best fit should be obvious
to you. If you had been presented with this set of data, your own line of best
fit might have been slightly different from the one shown.

81
Analysing and predicting costs

(4) This line can be used to predict the outcome at different levels of
activity.

In our example we are looking at the relationship between costs and activity
levels.

The point where the extrapolation of this line cuts the vertical axis (the
intercept) can be read off as the total fixed cost element. From the above
diagram the fixed cost contained within this set of data is adjudged to be
$200.

The variable cost per unit is given by the gradient of the line. The variable
cost is calculated as follows:

Cost for zero units = $200

Cost for 150 units = $500
500 – 200
Variable cost (gradient) = ————– = $2 per unit
150 – 0

Test your understanding 2

Based on the above scatter graph:

(a) the period fixed cost is $ .

(b) the variable cost per unit is $ .

82
 chapter 3

Advantages and disadvantages of scatter graph (line of best fit)

method

Advantages

• it is easy to understand and easy to use.

• all observations are used compared to the high­low method where only
two observations are used.

Disadvantages

• it relies on historical cost data and assumes this data can reliably
predict future costs.
• it assumes that the activity level is the only factor affecting costs.
• the accuracy of the line of best fit is questionable. Faced with the same
set of data, different lines may be drawn

5 Regression analysis

The third method which can be used to analyse and predict costs is
regression analysis.

The weakness of drawing a line of best fit ‘by eye’ should be obvious.

Regression analysis finds the line of best fit computationally rather than by
estimating the line on a scatter diagram. It seeks to minimise the distance
between each point and the regression line.

We know the equation of a straight line is y = a + bx. In regression analysis,

a and b are calculated as:

83
Analysing and predicting costs

If we are looking at analysing costs in relation to activity, the equation of the

line, y = a + bx would be made up of:

y = total cost

a = fixed cost

b = variable cost per unit

x = activity level

The high­low method effectively uses the same approach but only considers
the highest and lowest instances, which can lead to poor estimations. The
scatter graph method uses all past observations but there is judgement
used in drawing the line of best fit. Regression analysis is seen as a quicker
and more accurate method. The method used in the Fundamentals of
Management Accounting is least squares regression.

Illustration 3 – Regression analysis

An entity has the following data on its costs during the last year in each
of its regions and the corresponding number of units sold during this
time:

Region Cost ($000) Sales units (000)

A 236 11
B 234 12
C 298 18
D 250 15
E 246 13
F 202 10

Using least squares regression, calculate the fixed and variable

cost elements and present your answer in the form y = a + bx.

Solution

As we wish to forecast costs, we shall make this the dependent variable,

y, and the number of sales units the independent variable, x.

84
 chapter 3

The next step is to evaluate the parameters a (fixed cost) and b (variable
cost per unit). It is best to use columns to calculate the figures required
by the formulae:

x y x2 xy
11 236 121 2,596
12 234 144 2,808
18 298 324 5,364
15 250 225 3,750
13 246 169 3,198
10 202 100 2,020
––––– ––––– ––––– –––––
79 1,466 1,083 19,736
––––– ––––– ––––– –––––

Thus

and so

a = 244.33 – (10.12 × 13.17) = 111.05

Interpreting a and b

In this case, there is a fixed cost of $111,050 and a variable cost of

$10.12 per unit of sales.

If this was drawn on a graph, the fixed cost (a) would be the point of
intercept of the y axis and the variable cost per unit (b) would be the
gradient of the line.

85
Analysing and predicting costs

We can use this data to formulate the equation of the regression line in
the form y= a + bx:

y = 111.05 + 10.12x

Comparing least squares regression and scatter graph

From Illustration 3 above, the scatter graph has been drawn. The line of
best fit, drawn ‘by eye’ is shown in the top graph. In the bottom graph the
line has been drawn, for the same data, using least squares regression.

You should be able to see that the two lines are quite different and will
give different results.

This demonstrates that the scatter graph gives an approximation but

least squares regression gives a better, more accurate representation.

Note, for convenience, the scales on the axes do not start from zero.

86
 chapter 3

Test your understanding 3

A small supermarket chain has 6 shops. Each shop advertises in their

local newspapers and the marketing director is interested in the
relationship between the amount that they spend on advertising and the
sales revenue that they achieve. She has collated the following
information for the 6 shops for the previous year:

Shop Advertising expenditure Sales revenue

$000 $000
1 80 730
2 60 610
3 120 880
4 90 750
5 70 650
6 30 430

She has further performed some calculations for a linear regression

calculation as follows:

• the sum of the advertising expenditure (x) column is 450

• the sum of the sales revenue (y) column is 4,050
• when the two columns are multiplied together and summed (xy) the
total is 326,500
• when the advertising expenditure is squared (x2) and summed, the
total is 38,300, and
• when the sales revenue is squared (y2) and summed, the total is
2,849,300

Using regression analysis, establish the relationship between

sales revenue and advertising expenditure and present your
answer in the form y = a + bx, where sales revenue is y and
advertising expenditure is x.

87
Analysing and predicting costs

Advantages and disadvantages of least squares regression method

Advantages

• provides a more accurate estimation of the relationship between two

sets of data than other methods.
• information required to complete the linear regression calculations
should be readily available.
• computer spreadsheet programmes often have a function that will
calculate the relationship between two sets of data.

Disadvantages

• it relies on historical cost data and assumes that the historical

behaviour of the data continues into the foreseeable future.
• assumes a linear relationship between the variables.
• only measures the relationship between two variables. In reality the
dependent variable is affected by many independent variables.
• only interpolated forecasts tend to be reliable. The equation should not
be used for extrapolation.

6 Correlation

The reliability of the analysis we have carried out depends on the strength of
the relationship between the two variables. In our line of best fit and least
squares regression examples we have assumed a strong linear relationship
but this may not always be the case. The strength of the relationship
between variables is known as correlation.

Two variables are said to be correlated if they are related to one another,
or, more precisely, if changes in the value of one tend to accompany
changes in the other.

88
 chapter 3

A scatter diagram can reveal a range of different types and degrees of

correlation:

Graphs (a) and (b) show perfect correlation, i.e. there is a perfect linear
relationship between the variables. In (a) the relationship is positive which
suggests that y changes in line with x. In (b) the relationship is negative
suggesting that as y increases, x falls.

Graph (c) suggests a strong but not perfect positive correlation as not all
points on the graph would sit directly on the line of best fit.

Graph (d) shows a negative correlation. The relationship shown in (d) is less
strong than that shown in (c) as the points would be further from the line of
best fit.

Graph (e) shows no correlation suggesting that variable y has no relation to

variable x. In this case it would be impossible to draw a meaningful line of
best fit.

89
Analysing and predicting costs

Graph (f) shows a clear positive correlation which is not linear. This is known
as a curvilinear relationship.

The above shows that care must be taken in making predictions based on
historical data. The methods we have used in this chapter have assumed
perfect linear relationships. Where this is not the case, the predictions may
not be able to be relied on.

It is therefore important to measure the strength of the correlation between

the variables when using these methods. This can be done by calculating
the correlation coefficient.

7 Correlation coefficient

The statistician Pearson developed a measure of the amount of linear

correlation present in a set of pairs of data. Pearson’s correlation
coefficient, denoted r, is defined as:

where x, y and n are the same as in the least squares regression formula:

x is the independent variable

y is the dependent variable

n is the number of data points.

This measure has the property of always lying in the range –1 to +1, where:

• r = +1 denotes perfect positive linear correlation (the data points lie

exactly on a straight line of positive gradient);
• r = –1 denotes perfect negative linear correlation (again the data
points lie on a straight line but with a negative gradient); and
• r = 0 denotes no linear correlation.

The strength of a correlation can be judged by its proximity to +1 or –1: the

nearer it is (and the further away from zero), the stronger is the linear
correlation. A common error is to believe that negative values of r cannot be
strong. They can be just as strong as positive values except that y is
decreasing as x increases.

90
 chapter 3

Illustration 4 – Pearson's correlation coefficient

Evaluate Pearson’s correlation coefficient for the data on sales

and advertising spend in the table below, and interpret its value.

Advertising expenditure Total sales in following

in month month
($000) ($000)
1.3 151.6
0.9 100.1
1.8 199.3
2.1 221.2
1.5 170.0

Solution

Sales will depend on the level of advertising expenditure, therefore

advertising expenditure is the independent variable (x) and sales is the
dependent variable (y). We have five observations, so n = 5.

Set out a table as follows:

x y x2 y2 xy
1.3 151.6 1.69 22,982.56 197.08
0.9 100.1 0.81 10,020.01 90.09
1.8 199.3 3.24 39,720.49 358.74
2.1 221.2 4.41 48,929.44 464.52
1.5 170.0 2.25 28,900.00 255.00
––– –––– –––– –––––––– ––––––
7.6 842.2 12.40 150,552.50 1,365.43
––– –––– –––– –––––––– ––––––

Using

91
Analysing and predicting costs

Gives

The value of Pearson’s correlation coefficient in this case is 0.993. This

is close to 1, indicating a strong positive linear correlation.

Note: It is worth carrying out a simple check to ensure that your calculated
value for the correlation coefficient is between –1 and 1.

The arithmetic in such a calculation can be seen to be potentially very

tedious. In practice, spreadsheets can be used to do these calculations
quickly and accurately.

Test your understanding 4

A company owns six sales outlets in a certain city. The sales last year of
one of its key products (L) is given below, together with the sizes of each
outlet.

Floor space Sales of L

Outlet m2 '000 units
A 75 22.4
B 60 21.1
C 108 29.6
D 94 27.1
E 92 27.0
F 130 36.9

The company is investigating the effects of outlet size on sales.

Evaluate Pearson’s correlation coefficients for sales of L and the

size of outlets.

92
 chapter 3

Test your understanding 5

A company is forecasting costs for the next month and has been
analysing the data from the previous 11 months. It has established a
linear model showing how the level cost varies with the level of output.
Before relying on this model it wishes to ascertain the strength of the
relationship between costs (y) and output (x). The management
accountant has taken the following totals from a spreadsheet:

Σx = 440

Σy = 330

Σx2 = 17,986

Σy2 = 10,366

Σxy = 13,467

n = 11

Calculate the value of r, the coefficient of correlation, to two

decimal places.

A 0.98
B 0.63
C 0.96
D 0.59

8 The coefficient of determination

We have already seen how Pearson's correlation coefficient allows us to

discuss the strength of the relationship between two sets of figures.
However, the interpretation of the figure is made slightly easier if we square
the correlation coefficient, r, to give the coefficient of determination, r2.

The coefficient of determination, r2, gives the proportion of changes in y that

can be explained by changes in x, assuming a linear relationship between x
and y.

For example:

If a correlation coefficient r = +0.7, then r2 = 0.49 and we could state that

49% of the observed changes in y can be explained by the changes in x but
that 51% of the changes must be due to other factors.

93
Analysing and predicting costs

Test your understanding 6

If the correlation coefficient is 0.8, what is the coefficient of

determination?

A 0.64
B 89
C 20.8
D 0.4

9 Predicting costs

We have now looked in detail at the three methods we can use to analyse
semi variable costs between the fixed and variable elements. Once the
semi­variable cost has been split between fixed and variable elements, this
information can be used to forecast the total cost for another activity level
within the relevant range.

The possibility of changes occurring in cost behaviour patterns means that it

is unreliable to predict costs for activity levels which are outside the relevant
range. For example, our records might show the cost incurred at various
activity levels between 100 units and 5,000 units. We should therefore try to
avoid using this information as the basis for forecasting the level of cost
which would be incurred at an activity of, say, 6,000 units, which is outside
the relevant range.

Illustration 5 – Predicting future costs

Look back at the example from Illustration 1:

100 units 200 units

Material $500 $1,000
Labour $1,000 $2,000
Rent $2,000 $2,000
Electricity $700 $900

We worked out that material and labour were variable costs, rent was a
fixed cost and electricity was a semi­variable cost.

We can now use this information to predict the costs for a different level
of activity.

Estimate the total cost of producing 180 units.

94
 chapter 3

Solution

Material is a variable cost of $5 per unit. So the cost for 180 units will be
180 × $5 = $900.

Labour is a variable cost of $10 per unit. So the cost for 180 units will be
180 × $10 = $1,800.

Rent is a fixed cost of $2,000. This will be the same for 180 units.

Electricity is a semi­variable cost so to predict the cost for another level

of activity we first have to split it into its fixed and variable components
using the high–low method.

Change in costs
Variable cost per unit = ——————––
Change in units

900 – 700
Variable cost = ——–—— = $2.00 per unit
200 – 100

Fixed cost = Total cost – Variable cost

Using 100 units: Fixed cost = $700 – (100 × 2) = $500

Total cost = Total fixed cost + (Variable cost per unit × Number of units)

So the estimated cost of producing 180 units = $500 + ($2 × 180) =

$860.

The predicted total cost of producing 180 units will be:

$
Material 900
Labour 1,800
Rent 2,000
Electricity 860
———
Total 5,560
———

95
Analysing and predicting costs

Extrapolating outside the relevant range

In many examples you are told that the cost structures will remain
unaltered despite the increase in activity. In practice, you may need to do
an extrapolation outside the range for which you have available data.

Test your understanding 7

The variable production cost per unit of product B is $2 and the fixed
production overhead for a period is $4,000.

Calculate the total cost of producing 3,000 units of B in a period.

Test your understanding 8

A company has the following data on its profits and advertising

expenditure over the last 6 years:

Profits Advertising expenditure

$m $m
11.3 0.52
12.1 0.61
14.1 0.63
14.6 0.70
15.1 0.70
15.2 0.75

Using regression analysis, forecast the profits for next year if an

advertising budget of $800,000 is allocated.

10 Problems with using historical data to predict the future

In each of the methods considered in this chapter we have used historical

data to predict the future.

The main problem which arises in the determination of cost behaviour is that
the estimates are usually based on data collected in the past. Events in the
past may not be representative of the future and managers should be
aware of this if they are using the information for planning and decision
making purposes.

96
 chapter 3

11 Chapter summary

97
Analysing and predicting costs

12 End of chapter questions

Question 1

The finance manager of PQR is preparing the production budget for the
next period. Based on previous experience, she has found that there is a
linear relationship between production volume and production cost. The
following cost information has been collected in connection with
production:

Volume Cost
(units) ($)
1,600 23,200
2,500 25,000

Estimate the production cost for a production volume of 2,700

units?

A $5,400
B $25,400
C $27,000
D $39,150

Question 2

A company increases its activity within the relevant range. Tick the
correct boxes below to indicate the effect on costs.

Total variable costs will: increase o

decrease o
remain the same o
Total fixed cost will: increase o
decrease o
remain the same o
The variable cost per unit will: increase o
decrease o
remain the same o
The fixed cost per unit will: increase o
decrease o
remain the same o

98
 chapter 3

Question 3

The following data relate to two activity levels of an out­patient

department in a hospital:

Number of consultations per period 4,500 5,750

Overheads $269,750 $289,125

Fixed overheads are not affected by the number of consultations per

period.

Required:

The variable cost per consultation (to 2 dp) would be _________.

Question 4

The following data relate to the overhead costs of a commercial laundry

for the latest two periods.

Overhead cost Number of items

$ laundered
5,140 2,950
5,034 2,420

A formula that could be used to estimate the overhead costs for a

forthcoming period is:

Overhead cost = $ a + ($ b × number of items laundered)

Determine the value of a and b.

99
Analysing and predicting costs

Question 5

ABC is trying to understand the relationship between sales and

advertising expenditure. The management accountant has carried out
some analysis and has found that the coefficient of determination is
0.49.

Which of the following is correct?

A For every $0.49 spent on advertising, $1.00 of sales will be

generated.
B For every $1.00 spent on advertising, $0.49 of sales will be
generated.
C 49% of the variation in sales can be explained by the corresponding
variation in advertising.
D 49% of the variation in advertising can be explained by the
corresponding variation in sales.

Question 6

The following data relate to the cost of contract cleaners at two activity
levels:

Square metres cleaned 12,750 15,100

Cost $73,950 $83,585

Using the high­low method, what is the estimated cost if 16,200

square metres are to be cleaned?

A $88,095
B $89,674
C $93,960
D $98,095

100
 chapter 3

Question 7

Mr G has just opened 2 new stores in his local area. He knows if no

money is spent on advertising then sales will be $300,000, but for every
$1 spent on advertising sales revenue increases by $5. The predicted
advertising expenditure is expected to be $150,000 for one store and
$50,000 for the other.

Calculate the predicted total sales revenue for both stores.

Question 8

XYZ is investigating its current cost structure. An analysis of production

levels and costs over the first six months of the year has revealed the
following:

Production Production
Month level (units) cost
(000s) ($000)
January 9.0 240
February 10.0 278
March 9.7 256
April 10.5 258
May 11.0 290
June 11.5 300

Further analysis has produced the following data:

∑x = 61.7; ∑y = 1,622; ∑xy = 16,772; ∑x2 = 638.6

Using regression analysis, which of the following statements are

correct? Select all that apply.

A The variable cost per unit is $22.44

B The monthly fixed cost is $39.57
C The variable cost per unit is $39.57
D If 12,000 units were produced, the estimated total cost would be
$308,850
E The monthly fixed cost is $22.44
F If 12,000 units were produced, the estimated total cost would be
$474,862

101
Analysing and predicting costs

Question 9

Which of the following will adversely affect the reliability of a

regression forecast? Select all that apply.

A Small sample
B Low correlation
C Extrapolation
D Negative correlation

Question 10

The above graph plots total cost against activity levels.

Place the following elements on to the correct place on the

graph.

Cost $
Fixed cost
Variable cost
Activity units

102
 chapter 3

Test your understanding answers

Test your understanding 1

X is clearly a fixed cost as it does not change over the different activity
levels. For W, Y and Z it is not so clear, so calculate the cost per unit at
each activity level:

Cost type Cost per unit @ 100 units Cost per unit @ 140 units
$ $
W 80.00 (8,000/100) 75.43 (10,560/140)
Y 65.00 (6,500/100) 65.00 (9,100/140)
Z 67.00 (6,700/100) 61.29 (8,580/140)

The unit cost of Y is constant which suggests that Y is a variable cost.

The total cost and the cost per unit of W and Z vary at the different levels
which suggests that W and Z are semi­variable costs.

Test your understanding 2

The period fixed cost is $400.

The variable cost per unit is $3.

Change in costs
Variable cost per unit = ————————
Change in units

1,000 – 400
Variable cost = –———— = $3 per unit
200 – 0

Fixed cost = the intercept on the vertical axis = $400

103
Analysing and predicting costs

Test your understanding 3

b = (6 × 326,500) – (450 × 4050)

———–—————————
(6 × 38,300) – 202,500

136,500
= ———— = $5
27,300

4,050 450
a = ——— – 5 × ——
6 6

= 675 – 375

= $300

The relationship between sales revenue (y) and advertising expenditure

(x) is:

y = 300 + 5x

Test your understanding 4

Floor space is the independent variable (x) and sales is the dependent
variable (y).

x y x2 y2 xy
75 22.4 5,625 501.76 1,680
60 21.1 3,600 445.21 1,266
108 29.6 11,664 876.16 3,196.8
94 27.1 8,836 734.41 2,547.4
92 27.0 8,464 729.00 2,484
130 36.9 16,900 1,361.61 4,797
––– –––– ––––– ––––––– –––––––
559 164.1 55,089 4,648.15 15,971.2
––– –––– ––––– ––––––– –––––––

104
 chapter 3

The necessary summations are

• n=6
• Σx = 559
• Σy = 164.1
• Σx2 = 55,089
• Σy2 = 4,648.15
• Σxy = 15,971.2

Hence:

This is a very strong positive correlation between outlet size and sales of
L.

Test your understanding 5

Test your understanding 6

The coefficient of determination is given by squaring the correlation

coefficient.

Therefore r2 = 0.82 = 0.64. This suggests that 64% of the changes in y

can be explained by the changes in x.

105
Analysing and predicting costs

Test your understanding 7

Total cost for 3,000 units = $4,000 + ($2 × 3,000) = $10,000.

Test your understanding 8

First of all, to justify our assumption that there is a relationship between

the two variables, the correlation coefficient should be computed. It is left
as an exercise for you to verify that its value is 0.936. This high
correlation encourages us to proceed with the regression approach.

As we wish to forecast profits, we shall make this the dependent

variable, y, and advertising expenditure the independent variable, x.

The next step is to evaluate the parameters a and b:

x y x2 xy
0.52 11.3 0.2704 5.876
0.61 12.1 0.3721 7.381
0.63 14.1 0.3969 8.883
0.70 14.6 0.4900 10.220
0.70 15.1 0.4900 10.570
0.75 15.2 0.5625 11.400
–––– –––– –––––– ––––––
3.91 82.4 2.5819 54.330

Thus

and so:

a = 13.73 – (18.67 × 0.652) = 1.56

106
 chapter 3

The least­squares regression line relating profits to advertising

expenditure therefore has equation

y = 1.56 + 18.67x

Hence each extra million dollars of advertising generates an extra

$18.67 million profits. Also, profits would be $1.56 million without any
advertising.

If advertising expenditure is to be $800,000 (x = 0.8), then:

y = 1.56 + 18.67 × 0.8 = 16.496

Rounding this value off to a sensible level of apparent accuracy, we are

forecasting profits of $16.5 million next year, if advertising expenditure is
$800,000.

Question 1

Units $
2,500 25,000
1,600 23,200
——– ———
900 1,800
——– ———

$1,800
Variable cost per unit = ——— = $2
900

Substitute in high activity:

$
Total cost 25,000
Variable cost = 2,500 units × $2 5,000
———
Therefore fixed cost 20,000
———

107
Analysing and predicting costs

Forecast for 2,700 units:

$
Fixed cost 20,000
Variable cost 2,700 × $2 5,400
———
Total cost 25,400
———

Question 2

Total variable costs will increase

Total fixed cost will remain the same

The variable cost per unit will remain the same

The fixed cost per unit will decrease

Question 3

The variable cost per consultation would be $15.50.

With the same amount of fixed overheads at both activity levels, the
change in overheads must be due to extra variable cost.

Overheads Consultations
$
High 289,125 5,750
Low 269,750 4,500
———– ——–
Change 19,375 1,250
———– ——–

$19,375
Variable overhead cost per consultation = ———– = $15.50
1,250

108
 chapter 3

Question 4

a (fixed cost) = $4,550

b (variable cost per unit) = $0.20

Change in costs
Variable cost per unit = ————————
Change in units

5,140 – 5,034
Variable cost = —————— = $0.20 per unit
2,950 – 2,420

Fixed cost = Total cost – Variable cost

Using 2,950 units: Fixed cost = $5,140 – (2,950 × 0.20) = $4,550

A formula that could be used to estimate the overhead costs for a

forthcoming period is:

Overhead cost = $4,550 + ($0.20 × Number of items laundered)

Question 5

The coefficient of determination gives the percentage of the variation in y

(in this case, sales) which can be explained by the regression
relationship with x (in this case, advertising).

109
Analysing and predicting costs

Question 6

Change in cost
Variable cost per square metre = —————–—————
Change in square metres

Variable cost per square metre = ($83,585 – $73,950) ÷ (15,100 ­

12,750) = $4.10 per square metre

Substituting this back in to the data for 12,750 square metres we can
determine the amount of fixed cost:

Fixed cost = $73,950 – ($4.10 × 12,750) = $21,675

So for 16,200 square metres: $

Fixed cost 21,675
Variable cost (16,200 units × $4.10) 66,420
————
Total cost $88,095
————

Question 7

The equation of a straight line is y = a + bx. The information provided

shows that the sales revenue is dependent on the level of advertising.
The sales revenue is y and the money spent on advertising is x.

So the predicted total sales for the two stores is:

$
Store 1 (300,000 + 5 × 150,000) 1,050,000
Store 2 (300,000 + 5 × 50,000) 550,000
———–—
Total 1,600,000
———–—

110
 chapter 3

Question 8

A, B and D

A Variable cost = 6 × 16,772 – 61.7 × 1,622

————————————
6 × 638.6 – 61.72

554.6
= ——— = $22.44
24.71

1,622 61.7
B Fixed cost = ——— – 22.44 × ———
6 6

= 270.33 – 230.76

= $39.57

D The estimated cost of 12,000 units will be given by the linear cost
equation:

y = $39.57 + $22.44x

y = 39.57 + (22.44 × 12) = 308.85 = $308,850

Question 9

A, B and C

It is the strength of the correlation but not its sign that influences the
reliability of regression forecasts. Small samples, low correlation and
extrapolation all tend to give unreliable forecasts. Correlation can be
negative but still very strong so (D) is incorrect.

111
Analysing and predicting costs

Question 10

112