Adjusting Safety Stock Requirements

with an AHP-based Risk Analysis

Jukka Korpela Antti Lehmusvaara Kalevi Kyläheiko and

UPM-Kymmene Corp. South Carelia Polytechnic Markku Tuominen
00101 Helsinki, Finland Lappeenranta, Finland Lappeenranta University of
e-mail: jukka.korpela@upm- Technology, Finland
kymmene.com

safety stock to be practical and profitable [2]. The level of

Abstract customer service and safety stock is not an external
requirement, but it should be determined based on the
In most cases safety stock is determined based on the trade-off between the cost of providing the service and
variability of the demand and lead time. The focus has the contribution to revenues [3]. Very often companies
been on the statistics and past performance, and the are placing emphasis on reducing logistics and other
existing frameworks are not taking into account future costs to the lowest level possible to maximise the profit
development. Therefore we want to turn more to the potential of sales already on the books. Ideally, marketers
future issues and develop the existing frameworks by should be exploiting logistics capabilities to increase
implementing future factors to the decision making customer satisfaction and maintain customer demand [4].
process. In this paper, the safety stock requirements are Surprisingly, many companies are using outdated,
determined by applying a well-known decision support simplistic methods for allocating safety stock, and they
tool, the Analytic Hierarchy Process (AHP). While the don’t even realise that their system may be using
basis level for the safety stock is calculated by using the methods older than the electronic calculator [5]. They are
well-established calculation rules, the AHP-method is calculating safety stock based on old sales patterns,
used for adjusting this basis level by taking into account which means that they are not including future issues,
the risks related to e.g. production, logistics or the such as, company’s strategies and future intentions and
overall operating environment. By using the AHP, the risks related to business opportunities in their safety stock
risk factors are mapped in the various areas, their calculations.
importances are determined, and their impact on safety In some companies the determination of the safety
stock requirements is analysed. Furthermore, strategic stock level is based on sales forecasts. However,
viewpoints such as the strategies of the supplier company variations between actual demand and forecast are
and the customers can be included in the process of inevitable. A fundamental premise of forecasting is that a
determining the safety stock level. As the AHP is a forecast will be wrong, though we are not sure by how
flexible decision support tool, the safety stock much and when [6]. It’s not enough to take sales
requirements can easily be adjusted whenever there are forecasts to the determination of service level. Instead
changes in the risk or strategic factors. E.g. a strike companies have to consider all future issues in their
threat by the logistics service providers would trigger a safety stock calculations. All together, safety stock is a
safety stock requirement analysis. In the paper, the necessary strategic weapon of the business to enhance
utilisation of the AHP is demonstrated with an and maintain customer satisfaction and loyalty [6].
illustrative example. In our study service level is defined as the percentage
of orders that can be met from stock within a given
period. The elements of service level are the following:
1. The level of safety stock per product;
1. Introduction 2. The number of products belonging to the
safety stock.
An important element of customer service and
Provided that the sales history for a product forms a
logistics management is to maintain an appropriate level
routine pattern, the first can be calculated. Often, and
of safety stock. When insufficient safety stock is held, the
especially when the number of deliveries is high, the
firm fails to meet its customer service objective [1]. On
normal distribution function is used to determine the
the other hand, a 100% in-stock policy requires too much

Proceedings of the 36th Hawaii International Conference on System Sciences (HICSS’03)

0-7695-1874-5/03 $17.00 © 2002 IEEE
 needed safety stock. When the sales pattern follows the establish priorities among the elements within each level
normal distribution, the needed safety stock (S) can be of the hierarchy. The third step in using the AHP is to
defined when the value of its key parameters, safety stock synthesise the priorities of the elements to establish the
requirement (%), standard deviation of demand in time overall priorities for the decision alternatives. The AHP
unit (σ) and lead time (L) are known: differs from conventional decision analysis
methodologies by not requiring decision-makers to make
S = Z σ%L (1) numerical guesses as subjective judgements are easily
included in the process and the judgements can be made
entirely in a verbal mode [12].
This can be classified as a standard safety stock
According to Saaty [13], the Analytic Hierarchy
calculation, where Z is the Standard Normal Deviate and
Process forms a systematic framework for group
its value can be found on the basis of the service level
interaction and group decision making. Dyer and Forman
requirement from a standard normal distribution table
[14] describe the advantages of the AHP in a group
[7][8]. The problems related to the use of the function are
setting as follows: (1) both tangibles and intangibles,
due to the fact that it calculates safety stock based on the
individual values and shared values can be included in an
demand pattern and deviation of the past. It is not taking
AHP-based group decision process, (2) the discussion in
care of the future issues, such as strategic viewpoints of
a group can be focused on objectives rather than on
the deliverer and risks related to getting a deal with a
alternatives, (3) the discussion can be structured so that
customer.
every factor relevant to the decision is considered in turn,
The number of products belonging to safety stock can
and (4) in a structured analysis, the discussion continues
be determined by the ABC -analysis. For example, if 20
until all relevant information from each individual
per cent of the products (Class A) represent 70 per cent
member in the group has been considered and a
of the volume, then these class A products will establish
consensus choice of the decision alternative is achieved.
the safety stock and class B and C can be excluded.
A detailed discussion on conducting AHP-based group
The traditional method for defining the safety stock
decision making sessions including suggestions for
level relies very much on the historical performance of
assembling the group, constructing the hierarchy, getting
the supply chain, and it is difficult to incorporate future
the group to agree, inequalities of power, concealed or
development into the calculations. The objective of this
distorted preferences, and implementing the results can
paper is to present how the well-known decision support
be found in [15] and [13].
framework, the Analytic Hierarchy Process (AHP), can
In this paper, we propose the following approach to
be used for complementing the traditional approach. The
adjusting the safety stock level with the AHP:
AHP enables the decision-makers to take into account the
future development of the factors that have an impact on 1. Select the basic service level: The first step is
the supply chain performance. Furthermore, it is possible to determine the basic service level for a
to include the strategic viewpoints, like e.g. the strategies customer. The basic service level is expressed in
of the supplier and the customers, in the process of terms of a service level percentage (e.g. 99 %).
determining the safety stock level. 2. Calculate the safety stock requirement: The
basic safety stock requirement is calculated by
using the traditional calculation rule, i.e. on the
2. The proposed approach basis of the standard deviation of the demand
and the lead time.
The Analytic Hierarchy Process (AHP) is a theory of
3. Define the relevant factors with an impact on
measurement for dealing with tangible and intangible
the safety stock requirement: The third phase
criteria that has been applied to numerous areas, such as
of the approach involves determining the
decision theory and conflict resolution [9]. The AHP is a
relevant future or strategic factors that have an
problem-solving framework and a systematic procedure
impact on the safety stock requirement.
for representing the elements of any problem [10]. The
Examples of these factors are risks related to the
AHP is based on the following three principles:
supply chain (production, logistics etc.) and the
decomposition, comparative judgements, and the
future strategies of the supplier company and the
synthesis of priorities. The AHP starts by decomposing a
customer.
complex, multicriteria problem into a hierarchy where
4. Analyse and quantify the impact of the
each level consists of a few manageable elements which
factors with the AHP: The AHP is used for
are then decomposed into another set of elements [11].
analysing the impact of the defined factors on
The second step is to use a measurement methodology to
the safety stock requirement. With the AHP, the

Proceedings of the 36th Hawaii International Conference on System Sciences (HICSS’03)

0-7695-1874-5/03 $17.00 © 2002 IEEE
 factors are broken down into a hierarchy, the decision process is located at the first level, (2) the
elements in the hierarchy are prioritised and the criteria and the subcriteria at the following levels and (3)
overall impact on the safety stock level is the decision alternatives at the lowest level of the
determined. As a result of this phase, the impact hierarchy.
of each analysed factor is known as percentage In this analysis, the goal is to analyse the impact of
adjustment of the basic safety stock level that the customer B´s importance on the safety stock level.
was determined in the phase 2. The following criteria are included in the analysis: (1)
5. Summarising the results of the analyses: The profitability prospects of the customer B for the supplying
final step of the approach includes summarising company, (2) the impact of the customer B on the
the results of the analyses in the previous phase. supplying company, i.e. what the role of the customer B
This gives the overall need for adjusting the is in the customer portfolio of the supplying company in
safety stock requirement. the future, and (3) the nature of relationship between the
The utilisation of the proposed approach is customer B and the supplying company. These criteria
demonstrated with an illustrative, numerical example. are located at the second level of the hierarchy. The third
(i.e. the lowest level) of the hierarchy includes the
3. An illustrative example decision alternatives, which in this case are the
alternative adjustment percentages for the basic safety
In this illustrative example, we focus on the utilisation stock level. The hierarchy is presented in the Figure 1.
of the AHP, i.e. on the steps 4 and 5 of the proposed The described AHP-hierarchy can be considered an
approach. The basic set-up of the example, which covers application of the AHP to forecasting, and the structure
the steps 1-3 of the approach, is as follows: of the hierarchy is a simplified version of the one
• the supplier company A wants to determine the presented by Dyer and Forman [16] for forecasting
safety stock level for one of its major customers, the enrolment growth of certain educational programs.
customer B; i.e. the analysis is carried out for a Applications of the AHP for forecasting and prediction
single customer-product combination can be found in e.g. [13], [17], [18] and [19].
• the time frame is six months ahead The second step in using AHP is to derive priorities
• the basic service level requirement is 99 % and the for each element in the hierarchy. The priorities are set
corresponding safety stock level based on the by comparing each set of elements in a pairwise fashion
traditional calculation method is 2 000 stock keeping with respect to each of the elements in a higher level
units (SKUs) [11]. A verbal or a corresponding 9-point numerical scale
• the supplier company has defined three major factors can be used for the comparisons which can be based on
that need to be analysed with the AHP in order to objective, quantitative data or subjective, qualitative
adjust the safety stock level: (1) the importance of judgements. In a group setting, there are several ways of
the customer B – reflecting the future strategy of the including the views and judgements of each person in the
supplier company, (2) the estimated future priority setting process including e.g. (1) consensus, (2)
development of the customer B – reflecting the vote or compromise, (3) geometric mean of the
future strategy of the customer, and (3) the potential individuals’ judgements, and (4) separate models or
risks related to the supply chain during the next six players [14].
months
GOAL
TO ANALYSE THE IMPACT OF CUSTOMER
3.1 Analysing and quantifying the impact of the IMPORTANCE ON THE SAFETY STOCK
identified factors CRITERIA

PROFITABILITY IMPACT ON THESUPPLIER RELATIONSHIP

The impact of customer importance 0.196 0.493 0.311

The first of the factors to be analysed with the AHP is THE NEEDED
ADJUSTMENT
the importance of the customer B for the supplying -10- -5 % -5 - 0 % 0% 0-5 % 5-10 %

company. According to the basic principles of the AHP, Figure 1. The analysis of the impact of
the first step in the analysis is to break down the decision the customer importance
problem into components and to structure them into a
hierarchy. Normally, the basic structure of an AHP- With the hierarchy in the Figure 1, the priority setting
hierarchy is as follows: (1) the goal of the analysis or the process is started by comparing the criteria at the second

Proceedings of the 36th Hawaii International Conference on System Sciences (HICSS’03)

0-7695-1874-5/03 $17.00 © 2002 IEEE
 level pairwise with regard to the goal. The criteria are multiplication results (see e.g. [18]). The calculation of
compared to each other on the basis of their importance the composite adjustment rate is presented in Table 1.
in the process of determining customer importance. After As shown in Table 1, the composite adjustment rate is
finalising the pairwise comparisons of the criteria, the 0.05 %. The interpretation of this figure is the following:
alternative adjustment rates at the third level of the taking into account the future importance of the customer
hierarchy are compared pairwise separately with regard B to the supplying company A, the basic safety stock
to each criterion. For example, with regard to level should be increased by 0.05 %. Although the
profitability, the comparison process starts by asking the adjustment is only small, it shows that the customer B is
following: “With regard to profitability, which is more so important that the supplying company wants to secure
preferable – to adjust the basic safety stock level by 5-10 the customer service capability for this customer.
% or by 0-5 % -and how much more preferable is the Range Preference *
preferred alternative?” By following this procedure, the Adjustment Preference
average Range average
alternative adjustment rates are prioritised with regard to
each criterion. The overall priorities for the alternative 5-10 % 0,149 7,5 % 1,11 %
adjustment rates in the hierarchy are calculated by 0-5 % 0,267 2,5 % 0,67 %
summarising the priorities the adjustment rates have with
respect to each criterion. 0% 0,179 0% 0%
-5-0 % 0,262 -2,5 % -0,66 %
-10 - -5 % 0,143 -7,5 % -1,07 %
A TOTAL 1 0,05 %
-10- -5 % 14,3
D
J Table 1. The composite adjustment rate
5-10 % 14,9
U
S The impact of the future development of the customer
O% 17,9
T
M The second factor to be analysed is the future
-5 - 0 % 26,2
E development of the customer. The goal of the analysis is
N to determine the impact of the customer’s future business
T
0-5 % 26,7
development on the safety stock requirement. The criteria
that are used in this analysis are the following:
0,0 10,0 20,0 30,0
• stability of the customer’s business, i.e. will the
PREFERENCE (%) requirements of the customer be stable during the
next six months or can a significant increase or
Figure 2. The priorities of the alternative decrease be expected
adjustment rates • vulnerability of the customer to shortages of the
products supplied by the supplying company A, i.e.
The resulting priorities can be seen in the Figure 1. do the products of the supplying company have a
The most important criterion is the impact on the critical role in the production process of the
supplier company with a priority of 0.493. The priority of customer
the criterion “relationship” is 0.311 and that of the • supplier policy of the customer, i.e. what is the role
criterion “profitability” is 0.196. The overall priorities of of the supplying company A in the supplier portfolio
the alternative adjustment rates can be seen in Figure 2. (a sole supplier or one among many suppliers)
The results presented in Figure 2 show the overall • new requirements, i.e. will the customer B expand
preferences of the alternative adjustment rates with its business with new product launches during the
regard to the importance of the customer B. As shown in next six months resulting to increased requirements
the figure, the adjustment rate 0-5 % has the highest towards the supplying company A
preference. In order to calculate the composite The AHP-hierarchy has the same structure as the
adjustment of the safety stock level related to a customer hierarchy that was used for analysing the impact of
importance, the preferences of the alternative adjustment customer importance. The goal is located at the highest
rates are combined by multiplying the average of each level, the criteria at the second and the alternative
adjustment rate by its overall preference. The composite adjustment rates at the lowest level. The hierarchy is
adjustment is then defined by calculating the sum of the shown in Figure 3.

Proceedings of the 36th Hawaii International Conference on System Sciences (HICSS’03)

0-7695-1874-5/03 $17.00 © 2002 IEEE
 The priorities for the elements in the hierarchy are
derived by using the pairwise comparisons. The most
important criterion is “supplier policy”, which has a
priority of 0.340. The priorities of the alternative
adjustment rates of the basic safety stock level are shown
in Figure 4 and the calculation of the composite
adjustment rate is presented in Table 2. Based on the
analysis of the customer’s future development, the basic
safety stock level should be adjusted by –0.08 %.

The impact of supply chain-related risks

GOAL
TO ANALYSE THE IMPACT OF CUSTOMER´S
The third factor to be analysed is the impact of supply BUSINESS DEVELOPMENT ON THE SAFETY STOCK

chain-related risks. This analysis enables the supplying CRITERIA

company to take into account and predict future events STABILITY

0,237
VULNERABILITY
0,287
SUPPLIER POLICY
0,340
NEW REQUIR.
0,136
that could have an effect on the performance and
functionality of the supply chain. The following criteria THE NEEDED

are included in the analysis: ADJUSTMENT

-10- -5 % -5 - 0 % 0% 0-5 % 5-10 %

• production efficiency: This criterion refers to the

production capabilities of the supplying company Figure 3. The AHP-hierarchy for analysing the impact
taking into account e.g. the expected maintenance of the future development of the customer
and investment shutdown periods and the estimated
production efficiency (production rate per day)
during the next six months. A
• booking situation: The predicted booking situation D 5-10 % 14,5
of the production machinery has a significant impact J
on the supply chain performance. In a high-demand U -10- -5 % 15,2
season, the production is running full thus S
0-5 % 22,7
decreasing flexibility and increasing the risk for T
delays of customer deliveries. M
O% 23,6
• transportation reliability: The transportation part E
of the supply chain can be affected by multiple N -5 - 0 % 24,0
factors, such as predicted strikes. T
• transportation availability: This criterion takes into 0,0 10,0 20,0 30,0
account the availability of transportation capacity PREFERENCE (%)
which could become a problem especially during a
season of very high demand. Figure 4. The priorities of the adjustment rates
• warehousing: This criterion refers to all
warehousing-related factors with an impact on the Preference *
Range
supply chain performance. Adjustment Preference Range
The hierarchy structure and the priority deriving average
average
procedure are similar to the two previous analyses. The
AHP-hierarchy is shown in Figure 5, the priorities of the 5-10 % 0,145 7,5 % 1,09 %
adjustment rates in Figure 6 and the composite 0-5 % 0,227 2,5 % 0,57 %
adjustment rate in Table 3. The composite adjustment
0% 0,236 0% 0%
rate is 2.49 %, which shows that the supplying company
A expects to have some supply chain –related problems -5-0 % 0,240 -2,5 % -0,6 %
in the future and wants to secure the service level for the 0,152 -7,5 % -1,14
-10 - -5 %
customer B by increasing the basic safety stock level.
TOTAL 1 - 0,08 %

Table 2. The composite adjustment rate

Proceedings of the 36th Hawaii International Conference on System Sciences (HICSS’03)

0-7695-1874-5/03 $17.00 © 2002 IEEE
 GOAL
TO ANALYSE THE IMPACT OF SUPPLY CHAIN
RISKS ON THE SAFETY STOCK
CRITERIA

PRODUCTION BOOKING TRANSPORTATION TRANSPORTATION WAREHOUSING

EFFICIENCY SITUATION RELIABILITY AVAILABILITY 0,094
0,343 0,267 0,180 0,117

THE NEEDED
ADJUSTMENT -10- -5 % -5 - 0 % 0% 0-5 % 5-10 %

Figure 5. The AHP-hierarchy for analysing the impact of

supply chain –related risks 3.2 Summarising the results of the analyses

The last step of the proposed approach is to combine

the results of the AHP-analyses into one figure that
shows the overall needed adjustment of the basic safety
-10- -5 % 8,1 stock level. In this illustrative example, three AHP-
A analyses were conducted resulting to the following
D adjustment rates: (1) 0.05 %, (2) –0.08 %, and (3) 2.49
J -5-0 % 10,9 %. The simplest way to combine these results is to
U calculate their average which in this case would be 0.82
S %. This result means that the basic safety stock level
O% 18,2
T (2000 SKUs) of the customer B should be increased by
M 0.82 % on the basis of the customer importance analysis,
E the future development of the customer’s business and
5-10 % 27,0
N the analysis of the supply chain –related risks.
T However, the implicit assumption in calculating the
0-5 % 35,8 arithmetic average is that all the analyses have the same
importance. The AHP could be applied even in this step
to prioritise the analyses, as there might be differences
0,0 10,0 20,0 30,0 40,0
between the importance of the different factors. The
PREFERENCE (%) priorities of the analyses would then be used to calculate
the weighted average of the results of the analyses.
Figure 6. The priorities of the adjustment rates
4. Conclusions
Range Preference*
Adjustment
Adjustment Preference Traditionally, the safety stock level has been
average Range average determined based on statistics and past performance. The
proposed approach provides the decision-makers with
5-10 % 0,358 7,5 % 2,69 % improved possibilities to complement the traditional
0-5 % 0,270 2,5 % 0,68 % safety stock calculation process by quantifying the impact
of strategies and future events on the safety stock level.
0% 0,182 0% 0% The proposed approach results to an adjustment rate that
-5-0 % 0,109 -2,5 % -0,27 % tells the needed increase or decrease of the traditionally
calculated safety stock level.
-10 - -5 % 0,081 -7,5 % -0,61
The decision support tool used in this paper, the
TOTAL 1 2,49 % Analytic Hierarchy Process, provides a systematic and
flexible framework for analysing the impact of factors,
Table 3. The composite adjustment rate such as the importance of the customer, the future
development of the customer and the supply chain-

Proceedings of the 36th Hawaii International Conference on System Sciences (HICSS’03)

0-7695-1874-5/03 $17.00 © 2002 IEEE
 related risks. The flexibility of the AHP enables decision- [12] E.H. Forman, "Decision support for executive
makers to adapt the approach to match the specific decision makers", Information Strategy: The
circumstances of a certain situation. The AHP makes it Executive´s Journal (Summer), 1985, pp. 4-14.
possible to include both qualitative and quantitative [13] T.L. Saaty, Decision Making for Leaders, Lifetime
factors in the analysis as well as subjective judgements by Learning Publications, USA, 1982.
the decision-makers. An important aspect of the AHP is [14] R.F. Dyer and E.H. Forman, "Group Decision
that with it and the corresponding Expert Choice – Support with the Analytic Hierarchy Process",
software the analysis process can be documented in Decision Support Systems, Vol. 8, No. 2, 1992, pp.
detail. This also means that the analyses can easily be 99-124.
revised if e.g. new important information about a certain [15] B.L. Golden, E.A. Wasil and P.T. Harker, The
factor has been received. Analytic Hierarchy Process: Applications and
Studies, Springer Verlag, 1989.
References [16] R.F. Dyer and E.H. Forman, Analytic Approach to
Marketing Decisions, Prentice-Hall, Englewood
[1] W. Zinn and H. Marmorstein, "Comparing Two Cliffs, NJ, 1991.
Alternative Methods of Determining Safety Stock [17] T.L. Saaty and L.G. Vargas, Prediction, Projection
Levels: The Demand and Forecast Systems", and Forecasting, Kluwer Academic Publishers,
Journal of Business Logistics, Vol. 11, Issue 1, Dordrecht, 1991.
1990, pp. 95-110. [18] C. Wolfe, "How to Adjust Forecasts with the
[2] H. Meshkat and R.H. Ballou, "Warehouse Location Analytic Hierarchy Process", Journal of Business
with Uncertain Stock Availability", Journal of Forecasting, Vol. 7, 1988, pp. 13-17.
Business Logistics, Vol. 17, Issue 2, 1996, pp.197- [19] C. Wolfe and B. Flores, "Judgemental Adjustment
216. of Earnings Forecasts", Journal of Forecasting,
[3] P-K. Ho and J. Perl, "Warehouse Location under Vol. 9, 1990, pp. 389-405.
Service-sensitive Demand", Journal of Business
Logistics, Vol. 16, Issue 1, 1995, pp.133-162.
[4] T.P. Stank, P.J. Daugherty and A.E. Ellinger,
"Pulling Customers Closer through Logistics
Service", Business Horizons, Vol. 41, Issue 5, 1998,
pp.74-80.
[5] J.C. Sandvig, "Simple Solutions Aren’t the Best
Ones", IIE solutions, Vol.30, Issue 12, 1998, pp.28-
29.
[6] J.A.G. Krupp, "Managing Demand Variations with
Safety Stock", Journal of Business Forecasting
Methods & Systems, Summer 1997, Vol. 16, Issue
2, 1997, pp.8-13.
[7] R. Tersine, Production/Operations Management:
Consepts, Structure and Analysis, Second Edition,
North Holland, 1985.
[8] R. Tercine, Principles of Inventory and Materials
Management, Second Edition, North Holland,
1982.
[9] L.G. Vargas, "An Overview of the Analytic
Hierarchy Process and Its Applications", European
Journal of Operational Research, Vol. 48, 1990,
pp. 2-8.
[10] T.L. Saaty, "Priority Setting in Complex Problems",
IEEE Transactions on Engineering Management,
Vol. EM-30, No. 3, August, 1983, pp. 140-155.
[11] Y. Wind and T.L. Saaty, "Marketing Applications
of the Analytic Hierarchy Process", Management
Science, Vol. 26, No. 7, 1980, pp. 641-658.

Proceedings of the 36th Hawaii International Conference on System Sciences (HICSS’03)

0-7695-1874-5/03 $17.00 © 2002 IEEE