Developing a Multiple-Criteria Decision Methodology for the Make-or-Buy Problem

International journal of scientific and technical research in engineering (IJSTRE)
www.ijstre.com Volume 1 Issue 5 ǁ August 2016.
Manuscript id. 8166694 www.ijstre.com Page 1
Developing a Multiple-Criteria Decision Methodology for the
Make-or-Buy Problem
Ibrahim A. Alrammah
Mechanical, Aerospace and Civil Engineering (MACE) School, University of Manchester
United Kingdom
iarammah@gmail.com
Abstract: Engineering design is the process of developing a system, component or process to satisfy the desired
requirements. It is a decision making process, in which the basic mathematics and engineering disciplines are
utilized to convert resources optimally to achieve a predetermined objective. It also includes a variety of
realistic constraints such as reliability, safety, economic factors, ethical and social impacts. This work proposes
a methodology and a procedure for the make-or-buy problem. Companies following this methodology are
guided through a structured sequence comprising identification of factors for the make-or-buy decision, and the
comparison of internal sourcing and external sourcing factors against each other. Multi-attribute decision-
making is utilized to present an overall make-or-buy decision recommendation.
Keywords: Analytic Hierarchy Process (AHP), make-or-buy, multi-attribute decision-making
I. INTRODUCTION
This study ‎includes two stages. The first stage builds the conceptual model for the make-‎or-buy decision
problem. It develops the framework of the problem with a suggested procedure ‎that is composed of three
phases. The second stage contains a case of a hypothetical company ‎having a make-or-buy problem. The
solution includes quantitative and qualitative components. ‎Analytic Hierarchy Process (AHP) is used to
determine the importance of each criterion. Sensitivity ‎analysis and simulation tools are used which are: SensIt,
RiskSim and Excel spreadsheet. One advantage of the suggested methodology is that a spreadsheet based
optimizer is utilized to implement it. That it is one that provides optimal solutions and at the same time utilizes a
broadly used environment for implementation. So the suggested methodology should be a practical tool for
decision making. The ‎final decision for this case is optimized and recommendations are provided.
II. STATEMENT OF THE PROBLEM
The details of the case are as follows from Ref [1]: Planners at a hypothetical company are discussing the
decision whether to purchase a component that is considered as a part of the final product or to manufacture it
internally. “Forecasted annual demand for this part is 3200 units. The company works 250 days per year.
The company's financial specialists established a cost of capital of 14% for the utilization of funds for
investments inside the company. Furthermore, over the past year $600,000 was the average investment in the
inventory of the company. Accounting information displays that a total of $24,000 was spent on insurance
related to the inventory of the company. Also, an estimated $9000 was lost due to shrinkage of inventory, which
comprised damaged goods besides pilferage. A remaining $15,000 was spent on overhead of warehouse,
comprising utility expenses for lighting and heating.
An analysis of the purchasing operation displays that about two hours are required to process and
coordinate an order for the part regardless of the quantity of the order. Purchasing salaries average $28 per hour,
comprising employee benefits. Moreover, a detailed analysis of 125 orders revealed that $2375 was spent on
communication tools such as telephone, paper, and postage directly related to the ordering process. A one-week
lead time is necessary to get the part from the supplier. An analysis of demand during the lead time indicates it
is almost normally distributed with a mean of 64 units and a standard deviation of 10 units. Service level
guidelines show that one stock-out per year is satisfactory.
The company has the option to contract for purchasing the part from a supplier at random cost following
normal distribution with mean $13 per unit and standard deviation 3. However, over the past few months, the
company's production capacity has been expanded. Therefore, excess capacity now exists in certain production
departments, and the company is considering the option of producing the parts itself.
Forecasted utilization of equipment indicates that production capacity will be available for the part being
considered. The production capacity is available at the rate of 1000 units per month, with up to five months of
production time available. Management relies on that with a two-week lead time, schedules can be arranged so
that the part can be produced whenever required. The demand during the two-week lead time is almost normally
distributed, with a mean of 128 units and a standard deviation of 20 units. Production costs are estimated to be

Developing a Multiple-Criteria Decision ‎Methodology for the Make-or-Buy ‎Problem
$6 per part. An issue of management is that setup costs will be significant. The total cost of setup is
expected to be $50 per hour, and a full eight-hour shift will be required to set up the equipment for producing
the part. To produce the part in-house, 5 workers are needed with wage of $4 per hour. The work day is 8
hours” [1].
The question to be answered: Should the company purchase the part from the supplier or produce the part
itself?
III. MAKE-OR-BUY PROCESS
The make-or-buy process contains three phases:
Phase 1: Preparation
There are five areas for evaluating make-or-buy problem: (1) technology and manufacturing processes;
(2) supply chain management and logistics (SCM); (3) support systems; (4) social and ethics impacts; and (5)
costs. In this study, a quantitative area and a qualitative area will be picked, and a decision making
methodology will be applied. The two areas are: costs and SCM, and the factors affecting these areas in both
policies: "make" and "buy", will be examined. The scores of the two polices will be determined. The three
remaining areas will be given hypothetical scores, and the final decision is chosen according these scores.
Phase 2: Data Collection
Workshops were held and they defined the weighting of the relative importance of the various factors in
the framework and the rating of the performance of in-house and external supply [2]. AHP is used to find the
importance of each criterion. The problem is decomposed into a hierarchy of criteria and alternatives. In our
problem:
- The objective is to select the best policy
- The criteria are: (1) technology and manufacturing processes; (2) supply chain management and
logistics (SCM); (3) support systems; (4) social and ethics impacts; and (5) costs.
- The alternatives are: (1) make; (2) buy; (3) make some and buy some.
This information is then arranged in a hierarchical tree as in Fig.1.
Social Impact
Technology &
Manufacturing
Processes
Supply
Chain
Management
Support Systems
Make Or
Buy?
Make Buy
Make Some
& Buy Some
Fig.1: AHP for the make-or-buy problem.
We use judgments to determine the ranking of the criteria [3]. In this problem, the assigned team
decided that:
- Technology & Manufacturing Processes are 2 times important as SCM
- SCM is three times important as support systems
- and so on as shown in Table 1.
-
Table 1: The relative importance of the areas
Tech. & Manu.
Processes
SCM Support
Systems
Social
Impact
Tech. & Manu. Processes 1/1 2/1 1/5 4/1
SCM 1/2 1/1 3/1 5/2
Support Systems 5/1 1/3 1/1 8/3
Social Impact 1/4 2/5 3/8 1/1

After manipulations and matrix calculations, the computed eigenvector gives us the relative weights of
our criteria as follows:
Criteria Weights
Tech. & Manu. Processes 0.242
SCM 0.325
Support Systems 0.375
Social Impact 0.076
It is shown that “support systems” is the most important criterion, and the social impact is the least
important criterion.
Phase 3: Analysis and Results
In the third phase, a complete picture of the SCM and costs will be visualized. The ratings of supplier
and in-house would be determined based on costs calculation, while the area of SCM will be evaluated by a
questionnaire developed and scored by a team. After that, the findings of the weightings and ratings are
combined to present a single figure which gives an indication of the relative merits of making or buying options
[4].
IV. SUPPLY CHAIN MANAGEMENT ANALYSIS
In this section, the SCM which is a qualitative area will be examined. A team composed of four
members has met and agreed six factors affecting SCM. These factors are shown in Fig.2.
Fig.2 Factors of SCM area [5].
After setting these factors, the weightings have been determined. Each individual member in the team
has scored the factors for the two policies: “make” and “buy”. This process is summarized in Tables 2 and 3.
Table 2: Scores of “make” SCM.
Factor Score 1 Score 2 Score 3 Score 4 Score
Average
Weight Score
Weighted
Average
1 5 5 5 5 5 0.1 10
2 5 5 4 5 4.75 0.2 19
3 4 5 5 5 4.75 0.1 9.5
4 5 5 4 4 4.5 0.15 13.5
5 5 5 5 5 5 0.25 25
6 4 5 4 5 4.5 0.2 18
Total 1 95

Table 3: Scores of “buy” SCM.
Factor Score 1 Score 2 Score 3 Score 4 Score
Average
Weight Score
Weighted
Average
1 2 5 4 5 4 0.1 8
2 5 3 5 3 4 0.2 16
3 3 5 1 4 3.25 0.1 6.5
4 5 4 3 4 4 0.15 12
5 3 5 5 5 4.5 0.25 22.5
6 5 5 4 4 4.5 0.2 18
Total 1 83
We can see from Tables 2 and 3 that the “make” option is given 95 out of 100 score, and “buy” option is
given 83 out of 100 score. This means that the “make” policy is better than “buy” policy in terms of SCM.
V. COSTS ANALYSIS
The following factors will be considered in evaluating the alternatives:
1. An analysis of the holding costs, including the appropriate annual holding cost rate
2. An analysis of ordering costs for “Buy” policy, and an analysis of setup costs for the “Make” policy.
3. A development of the inventory policy for the following two alternatives:
 Ordering a fixed quantity Q from the supplier
 Ordering a fixed quantity Q from in-plant production
4. An analysis of the following in the policies of parts 3:
a. Optimal quantity Q*
b. Number of order or production runs per year
c. Cycle time
d. Reorder point
e. Amount of safety stock
Under probabilistic conditions, when demand and/or lead time vary, the reorder point often
includes safety stock.
f. Expected maximum inventory
g. Average inventory
h. Annual holding cost
i. Annual ordering cost
j. Annual cost of the units purchased or manufactured
k. Total annual cost of the purchase policy and the total annual cost of the production policy
[1].
5.1. Study of “Buy” Option
5.1.1. Economic Order Quantity (EOQ) Model
The assumptions of the EOQ model are as follow:
1) The daily demand rate, d, is constant and independent.
2) The order quantity, Q, is constant for each order and the entire order is received at one time.
3) The cost per order, Co, is constant and does not depend on the size of the order.
4) The unit cost, C, of the inventory item is constant and does not depend on the size of the order.
5) The inventory holding cost per unit per time period, Ch, is constant.
6) Shortages such as backorders and stock outs are not permitted.
7) The lead time for an order is constant. Lead time is the time between when an order is placed until it is
received.
8) The inventory level is reviewed on a continuous basis.
9) The planning horizon consists of multiple time-periods.
The inputs that we must consider in using the EOQ model to make the order quantity decision include: the
annual demand (D), number of days in the year, order lead time (m), unit cost (C), holding cost rate (I), and the
order cost (Co).
The holding cost rate, Ch, is used to compute the holding cost component.
The order cost, Co, covers preparation expenses for the order, such as payment, communication, invoice
verification, receiving and so on.

The main output variable is the order quantity, Q, how much to order every time an order is placed. This can
be determined by the EOQ model.
The EOQ model is derived as:
h
o
C
DC
QEOC
2
: *

Total Inventory Costs = Holding Cost + Ordering Cost
Reorder point (r) = m d where:
r = reorder point; d = demand per day; m = lead time for a new order in days.
Reorder Point under Conditions of Uncertainty
If normal distribution is used for lead-time demand, the general expression for reorder point, r:
Reorder Point = Mean of Lead Time Demand + Safety Stock
r = μ + z σ where:
μ is mean of demand during lead time
σ is standard deviation of demand during lead time
z is the number of standard deviations necessary to obtain the acceptable stock-out probability
z σ is the safety stock[1]
5.1.2. Solution of EOQ Model:
We will follow the procedure suggested earlier to examine “Buy” policy:
1) The holding cost is made of four items, as shown below. Note that those items are converted into
percent for a common unit of measurement. The holding cost applies to both making and buying
options:
o Cost of Capital = 14%
o Taxes/Insurance = $24,000 / $600,000 = 4%
o Shrinkage = $9,000 / $600,000 = 1.5%
o Warehouse Overhead = $15,000 / $600,000 = 2.5%
Total = Annual Holding Cost Rate (I) = 22%
Supplier purchase unit cost (C) = $13
Annual Unit Holding Cost Ch = IC = (.22)(13) = $2.86
2) The ordering cost applies to the "buy" option:
2 Hours at $28.00 = $56.00
Other Expenses = $2,375 / 125 = $19.00 [1]
Unit Order Cost Co = Total = $56.00 + $19.00 = $75.00
3) Formulas used to find the components of EOQ model are summarized in Table 4
Table 4: Formulas of EOQ model.
Description Formula
a Economic order quantity
h
o
C
DC
QEOC
2
: *

b Number of orders per year N = D / Q
c Cycle time 250 / N
d Reorder point r = μ + z σ
e Amount of safety stock z σ
f Expected maximum inventory Q* + z σ
g Average inventory 0.5 Q* + z σ
h Annual holding cost 0.5 Q Ch+ z σ Ch
i Annual ordering cost (D / Q) Co
j Annual cost of the units purchased C D
k Total annual cost of the “Buy” policy Ordering cost + holding cost
l Annual unit holding cost Ch= IC
m Probability of stock out #stock out per year / #orders per year

Fig.3: Summary of EOQ Results.
It is found from Fig.3 that the total annual cost of the “Buy” policy = $42,804.
Now the target to examine the case of “Buy” policy in which the annual demand (D), unit cost (C), holding cost
rate (I) and the order cost (Co) are varying within certain range. This task is done with aid of SensIt.
5.1.3. Sensitivity Analysis Using SensIt
As mentioned earlier, the company is deciding whether to manufacture a part or to purchase it from a
supplier. Using SensIt’s Many Inputs, One Output option to see how the model’s output depends on ranges
specified for each of the model’s input variables. Before utilizing this option, the model input cells are arranged

in adjacent cells in a single column, corresponding labels in adjacent cells are arranged in a single column, Low,
Base, and High input values for each input variable are arranged in three separate columns. Alternatively, the
three columns containing input values can be worst case, likely case, and best case. A proper arrangement is
shown in Fig.4 [6].
Fig.4: Model Display with Lower and Upper Bounds for "Buy" Policy in SensIt.
Fig.5: SensIt Tornado Numerical and Chart Output for "Buy" Policy
The uncertainty about Supplier Purchase Unit Cost (the topmost bar in the tornado chart) is associated
with the widest swing in Annual cost. For each Low output value in column E, the corresponding input value is
shown in column B. For each High output value in column G, the corresponding input value is shown in column
D. The tornado chart shows single-factor sensitivity analysis, i.e., for each output value, only one input value is
changed from its base case value [6].

Fig.6: SensIt Spider Numerical and Chart Output for "Buy" Policy
On a spider chart, lines that are nearly horizontal generally indicate an input variable where small
percentage changes do not have much effect on the output value. Lines that are more vertical indicate an input
variable where small percentage changes have a greater affect on the output value [6].
We can conclude from SensIt charts that the most affecting factors on cost for "buy" policy are: Supplier
Purchase Unit Cost (C) and Forecasted Unit Demand (D). From this discussion, we found that the Supplier
Purchase Unit Cost (C) is the most critical factor that affects the “Buy” choice. So, the variability of market
price may negatively affect the total cost and this might not be suitable for risk averse. This stage of (SensIt)
determines the critical inputs for the next stage (RiskSim).
5.1.4. Monte Carlo Simulation Using RiskSim
We can see in the summary output of RiskSim that the annual total cost mean of “Buy” policy is
$42,886. This result is close to the one we have obtained by the former solution, $42,804. However, the
solution obtained by RiskSim is more reliable in process of Make-or-Buy decision, since it takes into
consideration the randomness of demand and material cost, which is the case in real life.
Fig.7: One-Output Model Display for "Buy" Policy in RiskSim
Fig.8: RiskSim Numerical Output for One-Output of "Buy" Policy

Fig.9: RiskSim Summary Output for One-Output for "Buy" Policy.
5.2. Study of “Make” Option
In contrast with "Bay" option, we do not have a model like EOQ for probabilistic demand in production
case. Fortunately, simulation can be useful tool for getting reasonable solution for this problem. In our
simulation model, we want to find:
1) Q: the size of production run
2) T1: uptime (production time)
3) T2: downtime
4) T: the cycle length, where: T = T1 + T2
5) Annual setup cost
6) Annual holding cost
7) Annual production cost
8) Annual labor cost
9) Total annual cost
Fig.10 shows the Excel spreadsheet we developed for production scenarios simulation.

Fig.10: Excel spreadsheet simulation.
In the model presented in Fig.10, column A represents the (days) of simulation, we run the simulation for
20,000 days, i.e. 80 years, considering one year has 250 working days. Column B shows the number of units
produced every day, zeros represent days with no production (downtime). Column C shows the demand for
product every day which is normally distributed with mean 12.8 and standard deviation 2. Column D represents
the on hand inventory, which is inventory of the previous day plus production of the current day minus the
demand of the current day. Column E represents the stock-outs. If 1 is present, it is an indicator that negative
inventory happened in the neighbor cell in column D. Shortages are not lost but they are backlogged. If setup
has occurred, 1 is present in column F. Column G shows the daily holding costs, it equals (0.00088) * (6) * (on
hand inventory). We get the daily holding cost rate by dividing the annual rate by number of days in a year:
0.22 / 250 = 0.00088. The unit production cost is $6. Cell H2 shows the annual production cost, which is the
number of parts produced in a year times the production cost. Whenever there is production, there is labor cost
in column I which equals: 8 hours * 5 workers * $4 = $160. Cell E2 shows the total number of stock-outs
during the simulation period which is 80 years. Since it is allowed for only one stock-out in a year, an indicator
in cell E1 shows the word "true" if the total number of stock-outs in 80 years is less than 81, otherwise, the word
"false" appears. Cell K2 shows the annual total cost which is the summation of: setup, holding, production and
labor costs.
Note that every time we click on the spreadsheet, the values change, due the random numbers generated
by Excel. The value of total cost changes accordingly, but with low variation. As we increase the simulation
duration, i.e. 100,000 rather than 20,000 days, as we get more stable total cost. I tried several scenarios for Q, T1
and T2 to find the best scenario. Table 5 summarizes some of these scenarios.
Table 5: Scenarios for Q, T1 and T2
Description Q
(units)
T1
(days)
T2
(days)
Total
Cost
Note
1 Continuous production in
full capacity (5000 units in
a year) without downtime
20 25 0 $180,853 Very high total cost since we
produce much more than
demand, but results also in
less stock-outs (zero)
2 Full capacity with
downtime
50 10 15 $160,473 High total cost, but the labor
cost is reduced
3 Partial capacity with
downtime
50 7 18 $54,730 Fairly low total cost, but we
can try better
downtime
46 7 18 $36,010 Low total cost, but we can try
better
downtime
50 65 185 $35,621 Best scenario in terms of total
cast
Fig.11 shows the production cycles and the inventory level of one year for the winning scenario, produce
50 units every day in uptime which is 65 days, and stop producing for 185 days. Fig.12 shows the inventory
level for the whole period of simulation, which is 80 years. We notice that the inventory level keeps growing,
which means very high (infinite) inventory for infinite horizon. However, we can accept this scenario since the
simulation is run for 80 years which is long period, and the companies usually change their objectives and
regulations in shorter periods, i.e. 10 to 15 years, or even shorter.

Fig.11: Production cycles and the inventory level of one year
By comparing the two choices: "make" and "buy", we found that the "Buy" policy results in total annual
cost of $42,886, while the "make" policy results in total annual cost of $35,621. So, according to "cost" factor
in evaluating the "make" and "buy" options, "make" policy is better since it results in lower cost. However, the
third option is to “Make Some and Buy Some”. We will discuss this in the following section.
Fig.12: Inventory level for the whole period of simulation
5.3. Study of “Make Some and Buy Some” Option
We have discussed two options: to “buy” 100% of the quantity required to satisfy the demand, or to
“make” 100% of the quantity in-house. However, the optimal cost may exist between the two options. That is,
we will examine the option in which we “make” some of the quantity required and “buy” the rest. Table 6
summarizes several Scenarios of this policy.
Table 6: Scenarios for “make some and buy some” policy.
Percentage of
buy Percentage of make Buy cost
Make
cost Total Cost
0% 100% 0 35621 35621
10% 90% 4563 30327 34890
20% 80% 8879 24371 33250
30% 70% 13154 22930 36084
40% 60% 17416 21038 38454
50% 50% 21661 16264 37925
60% 40% 25903 14426 40329
70% 30% 30133 11141 41274
80% 20% 34363 6994 41357
90% 10% 38584 3629 42213
100% 0% 42886 0 42886
It is found from Fig.13 that the minimum cost exists when 20% of the quantity is bought and 80% is
made in-house. So, this policy is the best in terms of cost.

Fig.13: Costs vs. percentage of “buy”.
VI. SENSITIVITY ANALYSIS AND RESULTS
We did analysis on SCM, which is a qualitative area. We found that the “make” option got a score of 95
out of 100, and “buy” option got a score of 83 out of 100. For the rest of areas considered in process make-or-
buy decision, we assume that the designated team has met and decided the scores of each area.
Table 7: Areas Scores
Tech. & Manu.
Processes
SCM Support
Systems
Social
Impact
Overall
Score
Weight 24.0693% 32.4778% 35.7263% 7.6266 % 100%
“Make” Score 54 95 83 100 81.13076
“Buy” Score 89 83 64 61 75.89531
Gap -35 12 19 39
Table 7 shows the overall scores (81.13 for “make”, 75.89 for “buy”) showing that "make" is the
preferred option. However, not all areas support the "make" option to the same extent, and some area support
the "buy" option. It is in this situation where more detailed analysis needs to be undertaken to fully understand
and resolve any trade-offs. The first stage of this is a "gap analysis" showing the weighted differences for each
of the areas, this shows which areas contribute most to the overall score differences. In this case, "supply chain
management", "support systems" and "social impact" are positive. However, "technology & manufacturing
processes" is negative, indicating that this is the area of main weakness of internal manufacturing. The gap
analysis indicates relationships between changes in certain areas, either by the supplier or in-house, and the
effects these will have on the make-or- buy decision. This could be used to identify and put in place
improvement activities. Here, for example, if the indicated decision of "make" were adopted, the company
should also put in place activities to strengthen the "technology & manufacturing processes".
Although the main indicator for the preferred decision is the overall score, this single figure can be
misleading as it could be highly sensitive to particular factors. In order to assess this, sensitivity analyses are
performed, and graphs drawn to illustrate cross-over points which might indicate a change of recommendation
from make to buy.
Fig.14: Gap Analysis.
For the sensitivity analysis graph, two points are required. For the first point, a weighting value of zero
is given to the area under analysis, which evaluates the effect of not considering that area in the decision. The
weightings for the other areas are increased, maintaining their existing proportions, so that they sum to 100. The
overall scores are then calculated for both “make” and “buy”. For the second point, the individual scores for the
area under analysis are used, disregarding all the other areas. This evaluates the other extreme, considering only
the one area. The overall scores are calculated again. The points are then joined by straight lines. If the lines do
not cross, this means that the decision is not sensitive to changes in weight in that particular area. If the lines do
cross, the point where the lines intersect indicates the breakeven point; by increasing/decreasing the weight
above/below this point the decision is reversed. If the distance between the original weight and the point where
the lines cross is large, the decision is still not sensitive to changes in weight in that particular area. If the
distance between them is small, then the decision is sensitive to changes in weight in that area and a review of
weightings might be required.
For instance, in Fig.15, the sensitivity graph of "technology and manufacturing processes", the analysis
shows that if the weight of "technology and manufacturing processes" increased above the breakeven point,
25%, buying alternative becomes more attractive. Similarly, if the weight of "technology and manufacturing
processes" is less than 25%, making alternative becomes more attractive. Since the distance between the

original weight, 24% and the point where the lines cross, 25%, is short, the decision is sensitive to changes in
weight in area of "technology and manufacturing processes".
However, in Fig.16, "Sensitivity analysis for SCM", the graph has negative breakeven point, so, it is not
sensitive to weighting. In this case, the gap between the allocated weights and the breakeven points for the
sensitive factors was considered large enough for the outcome of the process to be robust. In Fig.17, “Sensitivity
analysis for support systems”, the graph has a breakeven point at weight of 35%, while the original weight is
36%. If “support systems” is given a weight greater than 35%, the “buy” policy becomes more attractive, and it
is given a weighting less than 35%, the “make” policy becomes more attractive. Since the distance between the
original weight and the point where the lines cross is short; the decision is sensitive to changes in weight in area
of "support systems".
Since the distance between the original weight and the breakeven point for “social impact” is large, the
decision is not sensitive to changes in weight this area.
Tech. & Manu. SCM Support Sys. Social Score
weight 0 0.428294 0.471132 0.100574 1
“Make” Score 54 95 83 100 49.16139
“Buy” Score 89 83 64 61 36.28749
Make Buy
0 49.16139 36.28749
100 54 89
Fig.15: Sensitivity analysis of "technology and manufacturing processes"
weight 0.356994 0 0.529889 0.113117 1
“Make” Score 54 95 83 100 74.57017
“Buy” Score 89 83 64 61 72.58549
Make Buy
0 74.57017 72.58549
100 95 83

Fig.16: Sensitivity analysis for “SCM”
weight 0.375065 0.506092 0 0.118843 1
“Make” Score 54 95 83 100 32.13781
“Buy” Score 89 83 64 61 40.6302
Make Buy
0 32.13781 40.6302
100 83 64
Fig.17: Sensitivity analysis for “support systems”
weight 0.260848 0.351974 0.387179 0 1
“Make” Score 54 95 83 100 46.22161
“Buy” Score 89 83 64 61 47.99488
Make Buy
0 46.22161 47.99488
100 100 61
Fig.18: Sensitivity analysis for “social impact”
Which policy to choose?

We see in Fig.19 that the scores of criteria is proportional to the percentage of “make”. That is, as we
make more and buy less as the benefits (criteria) scores increase. However, we see in Fig.20 that the minimum
cost exist when the benefits score is 82 which is corresponding in Fig.19 to the policy of “make 80% and buy
20%”. So, this is the winning policy that satisfies relatively high score of benefits and minimum cost.
Fig.19: Criteria vs. Alternatives
Fig.20: Costs vs. Benefits
VII. CONCLUSION
This project has satisfied the engineering design criteria by introducing a decision making process, in
which the basic math and engineering techniques are applied. Further, it has taken into consideration a variety
of realistic constraints such as: economic factors, level of technology, SCM and social impact.
In this work, a structured methodology is applied for make-or-buy problem on the case of a hypothetical
company. Going through the three phases of the process (a preparation phase; a data collection phase; and an
analysis and results phase) and after performing the analytical solution of the costing criteria, sensitivity analysis
and simulation were performed using SensIt, RiskSim and Excel spreadsheet. A qualitative analysis was
performed for SCM. It is found from the overall scores of making and buying that the company should make
the part in-house. It is concluded also from sensitivity analysis that the area of “technology and manufacturing
processes” and “support systems” are sensitive for weighting. So, more attention should be paid in evaluating
these areas. In spite of that the “make” is the winning policy in terms of the overall score; gap analysis shows
weakness in “Technology and Manufacturing Processes”. So, the company should also put in place activities to
strengthen this area.
In general, it is good idea to support “make” option, especially in the developing countries. By
enhancing this policy, the technology is transferred due to building new factories and getting new technical and
industrial experience. These benefits cannot be achieved by “buy” policy and treating the imported parts as a
“black box” without knowing their functions and internal structures. Another advantage of “make” policy is the
social impact. The recent studies has approved that the level of crime is proportional to unemployment.
Building new plants and factories offers more jobs for unemployed people, which decreases the level of crime
and enhance the prosperity of the society. This is one of the ethical issues in engineering. Even though the
“make” option may sometimes appear more costly than “buy” option, it might be less costly in the long term
and considering the other factors.
References
[1] D. R. Anderson et al.,An Introduction to Management Science(Thomson South-Western, 2008).
[2] ‎J.Padillo and M. Diaby. A multiple-criteria decision methodology ‎for the make-or-buy problem.
International Journal of Production Research, ‎Volume 37, No. 14, 1999, p. 3203- 3229.
[3] T.s Saaty, Fundamentals of Decision Making and Priority Theory with the Analytical Hierarchy
Process(RWS Publications, 1994).
[4] S.s J. Moschuris, Organizational participants in the make-or-buy process, Industrial Marketing
Management 37, 2008, p. 143–153.
Criteria vs. Alternatives
73
74
75
76
77
78
79
80
81
82
0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%
Percentage of "Make"
ScoresofCriteria
Costs vs. Benefits
0
5000
10000
15000
20000
25000
30000
35000
40000
45000
50000
75 76 77 78 79 80 81 82
Benefits
Costs

[5] K.W. Platts, D.R. Probert, L. Make vs. buy decisions: A process incorporating multi-attribute
decision-making. International Journal of Production Economics, Volume 77, Issue 3, 11 June 2002,
p. 247–257.
[6] M. R. Middleton, Decision Analysis Using Microsoft Excel(School of Business and Management,
University of San Francisco, 2006).

Developing a Multiple-Criteria Decision Methodology for the Make-or-Buy Problem

More Related Content

What's hot (20)

Similar to Developing a Multiple-Criteria Decision Methodology for the Make-or-Buy Problem (20)

More from International journal of scientific and technical research in engineering (IJSTRE) (20)

Recently uploaded (20)

Developing a Multiple-Criteria Decision Methodology for the Make-or-Buy Problem