ADVANCED MANAGEMENT ACCOUNTING

Lecture (6)

Linear Programming

Dr Owolabi Bakre

Cost Behaviour Summary

The use of Least-Squares Regression Method to analyse mixed costs and estimate cost functions. It is objective and incorporate all of the available observations into the cost estimate.
 Simple linear regression (manual and computer solutions Excel & SPSS)  Multiple linear regression  Non-linear regression (the learningcurve-effect)
Dr Owolabi Bakre 2

Briefly Reviewing Cost Classifications for Decision Making For decision-making classified into:
 Differential Cost,  Opportunity Cost,  Sunk Cost.

Dr Owolabi Bakre

Differential Costs and Revenues

Costs and revenues that differ among alternatives.
Example: You have a job paying 1,500 per month in your hometown. You have a job offer in a neighbouring city that pays 2,000 per month. The commuting cost to the city is 300 per month. Differential revenue is: 2,000 1,500 = 500 Differential cost is: 300

Dr Owolabi Bakre

Opportunity Costs
The potential benefit that is given up when one alternative is selected over another.
Example: If you were not attending university, you could be earning 15,000 per year. Your opportunity cost of attending university for one year is 15,000.

Dr Owolabi Bakre

Sunk Costs
Sunk costs cannot be changed by any decision. They are not differential costs and should be ignored when making decisions.
Example: Example: You bought a car that cost 10,000 two years ago. The 10,000 cost is sunk because whether you drive it, park it, trade it, or sell it, you cannot change the 10,000 cost.

Dr Owolabi Bakre

What is Linear Programming?

Managers are routinely faced with the problem of deciding how constrained resources are going to be utilised. There is an opportunity cost for scare resources that should be included in the relevant cost calculation for decision making. For example, a firm may have:  limited raw materials,  limited direct labour hours available, and  limited floor space. How should the firm proceed to find the right combination of products to produce? The proper combination or mix can be found by use of a quantitative method known as linear programming.
Dr Owolabi Bakre 7

What is Linear Programming? (cont)

Linear Programming (LP) is a powerful mathematical technique that can be applied to the problem of allocating constrained (scare) resources among many alternative uses in such a way that the optimum benefits (total contribution margin) can be derived from their utilisation in the short run. (Notice: fixed costs are usually unaffected by such choices).
Dr Owolabi Bakre 8

What is meant by a linear programming problem?

A linear programming problem consists of three parts:
 1) a linear function (the objective function) of decision variables (say, x1, x2, , xn) that is to be maximized or minimized.  2) a set of constraints (each of which must be a linear equality or linear inequality) that restrict the values that may be assumed by the decision variables.  3) the sign restrictions, which specify for each decision variable xj either (1) variable xj must be nonnegative xj >= 0; or (2) variable xj may be positive, zero, or negative xj is unrestricted in sign
Dr Owolabi Bakre 9

A single-resource constraint problem

Where a scare resource exists, that has alternative uses, the contribution per unit should be calculated for each of these uses. The available capacity for this resource is then allocated to the alternative uses on the basis of contribution per scare resource.
Dr Owolabi Bakre 10

Example from Seal et al. (2006), Chapter (9), pp. 366-367

Mountain Goat Cycles makes a line of panniers (saddlebags for bicycles). There are two models of panniers:
 A touring model, and  A mountain model

Cost and revenue data for the two models are given below:

Dr Owolabi Bakre

11

Mountain Goat Cycles Example (cont)

Model
Mountain Pannier Touring Pannier

Selling price per unit Variable cost per unit Contribution margin per unit Contribution margin (CM) ratio

25 10 15 60%

30 18 12 40%

According to this information, the mountain pannier appears to be much more profitable than the touring pannier.

Dr Owolabi Bakre

12

Mountain Goat Cycles Example (cont)

Now let us add one more piece of information
 the plant that makes panniers is operating at capacity.

The bottleneck (the constraint) is a particular stitching machine. The mountain pannier requires 2 minutes of stitching time, and each unit of the touring pannier requires one minute of the stitching time. In this situation, which product is more profitable?
Dr Owolabi Bakre 13

Mountain Goat Cycles Example (cont)

Model
Mountai n Pannier Touring Pannier

Contribution margin per unit (from above) (a) Time on stitching machine required to produce one unit (b) Contribution margin per unit of the constrained resource (c)= (a) (b)

15 2 7.50

12 1 12

According to this information, the touring model provides the larger contribution margin in relation to the constrained resource.

Dr Owolabi Bakre

14

Mountain Goat Cycles Example (cont)

To verify that the touring model is indeed the more profitable, suppose an additional hour of stitching time is available and there are unfilled orders for both products. The additional hour could be used to make either 30 mountain panniers or 60 touring panniers.
Dr Owolabi Bakre 15

Mountain Goat Cycles Example (cont)

Model
Mountain Pannier Contribution margin per unit (from above) (a) Additional units that can be processed in one hour Additional contribution margin Touring Pannier

15 X 30 450

12 X 60 720

This example clearly shows that the touring model provides the larger contribution margin in relation to the constrained resource.

Dr Owolabi Bakre

16

Then, the rule in the case of a single resource constraint problem

Should not necessarily promote those products that have the highest UNIT contribution margins. Total contribution margin will be maximised by promoting products or accepting orders that provide the highest unit contribution margin in relation to the constrained resource.
Dr Owolabi Bakre 17

Another Example from Drury (2004), Chapter (9), PP.321-322

Components Contribution per unit Machine hours per unit Estimated sales demand (units) Required machine hours Contribution per machine hour Ranking per machine hr X 12 6 2,000 12,000 2 3 Y 10 2 2,000 4,200 5 2 Z 6 1 2,000 2,000 6 1
18

Capacity for the period is restricted to 12 000 machine hours. Dr Owolabi Bakre

Another Example from Drury (2004), Chapter (9)- cont

Profits are maximized by allocating scarce capacity according to ranking per machine hour as follows: Production Machine hours Balance of machine used hours available

2,000 units of Z 2,000 units of Y 1,000 units of X

2,000 4,000 6,000

10,000 6,000 -

The production programme will result in the following: 2 000 units of Z at 6 per unit contribution 12 000 2 000 units of Y at 10 per unit contribution 20 000 1 000 units of X at 12 per unit contribution 12 000 Total contribution 44 000

Dr Owolabi Bakre

19

More than one scare resource problem

Graphical Method (two products + many constraints) Simplex Method (More than two products + many constraints)
 Manual (tables/ matrices)  Computer programs:
Lindo MS Excel

Dr Owolabi Bakre

20

Graphical Method
Any linear programming problem with only two variables (i.e. two products) can be solved graphically. We may graphically solve an LP (max problem) with two decision variables as follows: Step (1): Graph the feasible region. Step (2): Draw an iso-contribution line. Step (3) Move parallel to the iso-contribution in the direction of increasing the objective function. The last point in the feasible region that contactsan iso-profit line is an optimal solution to the LP.

Dr Owolabi Bakre

21

Example from Seal et al (2006), P.893

Colnebank Ltd is a medium-sized engineering company and is one of the large number of producers in a very competitive market. The company produces two products, pumps and fans that use similar raw materials and labour skills. The market price of a pump is 152 and that of a fan is 118. The resource requirements of producing one unit of each of the two products are: Material (kg) Labour hours Pump 10 22 Fan 15 8 Material costs are 4 per kg and labour costs are 3.50 per hour During the coming period the company will have the following resources available to it: 4,000 kg of materials and 6,000 labour hours Required: Determine the mix of products (pumps and fans) that will maximise Colnebanks profit for the coming period, using graphical solution.

Dr Owolabi Bakre

22

Solution:

formulating a mathematical model of the above problem

Pumps () Fans () 60 40 (10kg at 4)

77 (22hrs at 3.50) 28 (8 hrs at 3.50) 117 88 Selling price 152 118 Contribution 35 30 The linear programming problem is to maximise the total contribution subject to the constraints. If P= units of pumps produced and F= units of fans produced, then the problem may be set up as follows:
1. 2. 3. Maximise: 35P + 30F Subject to: 10P + 15F < 4000 (materials) 22P + 8F < 6000 (labour hours)

Materials (15kg at 4) Labour

Dr Owolabi Bakre

23

Graphing the feasible region

Dr Owolabi Bakre

24

Steps (2) & (3): Drawing an iso-contribution line and moving parallel to the iso-contribution in the direction of increasing the objective function

Dr Owolabi Bakre

25

Which is the best combination of products?

Dr Owolabi Bakre

26

Sensitivity Analysis
Sensitivity Analysis involves asking what-if questions. For example, What happens if the market price of pumps falls to 145? What will be the loss of contribution and will the revised contribution change the optimal combination?

Dr Owolabi Bakre

27

Sensitivity Analysis (cont)

Dr Owolabi Bakre

28

Shadow Prices
Each constraint will have an opportunity cost, which is the profit foregone by not having an additional unit of the resource. In linear programming, opportunity costs are known as shadow prices. Shadow prices are defined as the increase in value that would be created by having one additional unit of a scare resource. For example: What is the additional contribution if one extra labour hour is available?
Dr Owolabi Bakre 29

Shadow Prices (cont)

If one extra labour hour is obtained, the constraints 10 P + 15 F <= 4000 and 22 P + 8F <= 6000 will still be binding, and the new optimum solution can be determined by solving the following simultaneous equations: 10 P + 15 F <= 4000 (unchanged labour constraint) 22 P + 8 F <= 6001 (revised labour constraint) The revised optimal output when the above equations are solved is 111.96 units of F and 232.06 units of P. This will increase contribution by 0.90 (the shadow price of labour hours)

Dr Owolabi Bakre

30

Another Example from Drury (2004), P.1110

Dr Owolabi Bakre

31

Another Example from Drury (2004)- cont

Dr Owolabi Bakre

32

Materials constraint (8Y + 4Z 3,440 (When Y= 0, Z = 860; when Z= 0, Y = 430

Dr Owolabi Bakre

33

Labour constraint 6Y + 8Z 2,880 (When Z = 0, Y = 480; When Y = 0, Z =360)

Dr Owolabi Bakre

34

Machine capacity constraint 4Y + 6Z 2,760 (When Z = 0, Y = 690; when y = 0, Z = 460)

Dr Owolabi Bakre

35

Sales limitation Y 420

Dr Owolabi Bakre

36

Optimum solution

Dr Owolabi Bakre

37

Optimum solution

Dr Owolabi Bakre

38

Simplex Method: Manual Solution

In 1947, George Dantzig developed the simplex algorithm for solving linear programming problems. The simplex algorithm proceeds as follows:  Step (1): Convert the linear programming problem to standard form.  Step (2): Obtain a basic feasible solution (bfs), if possible, from the standard form.  Step (3): Determine whether the current bfs is optimal.  Step (4): If the current bfs is not optimal, then determine which non-basic variable should become a basic variable and which basic variable should become a non-basic variable to find a new bfs with a better objective function value.  Step (5): Find the new bfs with the better objective function value. Go back to step (3).
Dr Owolabi Bakre 39

Applying the simplex algorithm

See the Drurys example, p.1110

Dr Owolabi Bakre

40

Simplex method: Drurys example

Dr Owolabi Bakre

41

Simplex method contd.

Dr Owolabi Bakre

42

Simplex method contd.

Second Matrix

Dr Owolabi Bakre

43

Simplex method contd.

Dr Owolabi Bakre

44

Simplex method contd.

Dr Owolabi Bakre

45

Simplex method contd.

Dr Owolabi Bakre

46

Computer Programs
The use of the LINDO Computer Package to solve linear programming problems The use of Microsoft Excel to solve linear programming problems

Dr Owolabi Bakre

47

Assumptions underlying linear programming

1. 2. 3. 4. Linearity. Divisibility of products. Divisibility of resources. All of the available opportunities can be included in the model. 5. Fixed costs are constant for the period. 6. Objective of the firm (maximise short term contribution).
Dr Owolabi Bakre 48

The Limitations of Linear Programming

LP ignores marketing considerations. It has an extensive focus on the short term. Furthermore, most production resources can be varied even in the short term through overtime and buying-in. The alternative to optimizing against given constraints is to concentrate on managing constraints. The theory of constraints helps in managing constraints in the short term. It is the topic of the next lecture.

Dr Owolabi Bakre

49

Workshop (3)
See Exercises P21-4 & C21-8 (Seal et al., 2006)

Dr Owolabi Bakre

50