t

S
(L
:r:
c:(
I
ell c "
""CJ a
c 0 '€l a,
QJ 'B
:if " .2 ::c "
"t:I '" ~ ,., "
~ " "0 .2
~ "
" 0 ;; .,
"
n, a ~ ~ U
a. "
~ " ~ " a
if) o o
..: .... 0 -.t .,;
Ii
t.
0 . ':' .. "-:",:::',.:',".:':

[ •..• 111[' .•.• ;1 .• 11'.11

i ···:1· .... ·.·.·

.f,"it IljUftljjil1

!I.:.I ..•..••.•.

:,..
•
:.: In :!:!oo Gl II
c :::J .. w 1Il. 0
~ ~. ~ i,j' 5' !;!.
:::J ~~
:::J. 3 ~ 'i'
r; '" IX 3
:2. a ..
.. if
0 !!;1
<t> ....
0 2
.r 5'
'" ~
0 ii>
"
;;: Sl @
~ 5' Q,
::r .,
0 3 ru
... 0.
.. c
13 :Q. s
:::J '"
tr 0'
% :::J
'<3
'" 0 :,..
9.~ "'0 s •
~'" 2.
9'''' '" 0 l~
'" . alb' c.
0.<0 :::r ,,r 0 II
.,8 ",,, ~
Etc. g"
",:::J '" " ~t
~'" ",E!.
0.1:, <t>ti
:i' 9. ~ ~~
3 .,
o " "':il 0
~tf ~2. In
<t>"" ~,
:f a ~IX
.,'" ~g 0'
ga. ::J
"'~ "'''' ?;
o.ro ~'"
<t> 3 ~$ ru
§:flo " .:z
o <: ~9. I!?,
:::J'" fA
~. 0.. .,
E'''' OJ
a~
(5' a. '"
0
? '" :::J
0 E!.
:::J ."
g a
o g
~ "
c "
Ol it
~ <'-
..9.!
a. ..
a. "
ro g
"'s
..9.!o c
'" (i~~ tl
OJ
! ~~g 0
.... E"
c a. ~~t Oro
4) '" 5E
E j c:r..,e> .,
l!! ro tl <II 3l ~
:;, C> :E~; -8.$
<II
m c 8,': ~
·c z,~
E :1 .... t::._ ~8
4) E .!2~:;:;
> 8.'.1 ):: ~ 1l.~ 21'"
"" ~ §
.!S! ~~ OJ ",.,.Q
4) "'tl OJ ._ III ., tl <1J
a:: <1J;'< ~ ....... <J.> .0""
g~~ ~ Ol
~~ .!:l " E
Q.i'~ <11 ~~~ E,S
i5fi 0.0 a. '" <1J
~i E- o, O<1J£; j!o.
~l ro E:S ~ 0.0
~I s: o..r::;
0 ~~~ Ol tl
:5 ~~ § :E <11 >:
• ~ :c ~:'!::::
.... :t:0l .
• ~
"" .0
E
.,; e> .S!
0 s e
= ~
., a.
:;; ~
S '5 'l'i.
" ~ E
:;;!
'" .~ 8
o- :;; o
a. c .<!: .<!:
~ ~ o " s
'M ::J
§ 'g ;£. 2
.':2 Q '" '"
ro e .~ '" Q.
0- if) OJ
.c ~ ., _j .r::;
:s 1: '" .,
e o § =
~ <1J '0
~ n. 2
'" F
0 :; .,
e ,., '"
ofi :f, .D "
OJ 2 al 2
h ~ Q Q. .,
U o 1 n.
Q. :2 E
:r: c: 'to
• -c -c Q -c
.... . .
. o~

Ij

• .....

•

'<Ii ~
E
'"
'"
k
,g
..,
..
.., '"
'" .~ "
.. »
o
c c: 'iil
.. i: .., -o
0 " '" .~
"" CT '" .1'2
o ., ., ....
Q) <J) '0
&i c '0 0 8
0 ., ~
o .1'2
.~ '" c e
" '" E 0..
a. 0 0 '" c-,
0.. .~ o ·c
:J 'E ~ ~ ~~
(f) ~ '" E ,g r1~
0 '" ~ ~~
a. .c BID :;
.!; .~ 0.<0 E o~
II ... 0. ",'" '" ~'iii
"
0 a. .~ ~ '0 'iii '0
'0 "
c .., .~ 5 t:: ~~
.. '" ~~ 8.
E c .c
~ a. o..E
1. ~ ~g " :r:"
<f) <>:c
. l
Ii
:;-
8 m
" ;u ~
V> zs (I) -n 0
!Ii -c it if a,
'" a" g '" s
" is' §: ::J.
::J ~ .. ::I .
~ 3 ::J .w ::J 0
'" 0 8 ::I
;u a .. o
., ~
<7.
0 ::l.
!ll
~
6'
~ ;T
""
0 0 0 0 a a ::J
" " ~ ;.., " u. e '"
'" .. ~ ~ OJ s:
0 <.n ,r
'" II.
~ :'S :'S ~ a '" '" t::
:!! ~ ~ so
OJ 0 "i "i "! "!
c "
32 til
c
r:!
~
'a..
a.
:J
(J) ~ U .,; to W 0 u, (,9
~ ~ k ~ k _9;! k
" i '" ~
0. a li i5.
t>- o. 0. c,
::J o, o, 0. 0. "'- "'- o,
til " " ::J ::J " " "
if) if) if) if) C/l if) til
g~
II
•
...
• • ...

.

00

/

/

. . :t . . ;.
~ .. ~ ~ -,! '" s:
0;' ::T <: •
='l :::J "[] <> :g "!J ,~
3 S ;§ B: ~ ~ 0;'
<: _, ~ " (il <: 8
~ 3 <1> " <l
16% 3 fir ~ ~~
'"
3 <1>
~, '" ., C'
:::J '0 S-
It) It "[] ::>
a c. a s
.,
5 ::i' g. .,
c 5' "[]
... :t- '!2.
'" It) :t Ei' 0
go f "!J 0
'" en C' :::J
0 " o.
3 :cl i5' c
'" g .,
go " ~ 1!2,
0 0 a. 0
a. :::J s: Oi :::J
~ .a
'U o
., "
3 ~ c
3
15 If
~ Q,
.,
l:l
I;l i jj ~
ii {l
"" ~ .S
~ S ..
" 0. ~
~ ~ ~ :i
s
.. c-, S .Y1
,::; N S ] "
0 =
'.p a '" l .~
(U " ~ a
"
> ] ] c-, f
'.p "
0 -e ..c
:2: ~ 0. § Z g
"
c x <i :.s
~ x
U ~ 1 E
~ ~
~ c.
~ 0. B
~ ",!Ii 8 e-, 8
s-, 0
·1 "" -o .E .., .S
j\:; x .5 " " ~
..,
'" ,..; 0; <" f-<
Ii
u
•
--
• <Il lJ c Q)

:t

j)

•

- .

•

(1 ~ i'>
!) ~. g:' ""> fJ~
i!~o is
2..'<:10 ~~.
ij:[ g. !t B
~ ...
6- ~ ;; t~ ~~
" e- = ~~
~ '" a~
~!l o[ s
~ :s El s- ~
~~ '"0'" ~
~a r-
~ g. '"
-g. l
""= ;0."
0<> ~ s·
tl0 OJ " ~
'"o;J ,," t
§"g ". a
-" s j
10
a 0
R 5' 0 lD ?>
o a"- t/) i
E!. I
~ N
II "
§. "
g l! s, G
g, s, ~ ~.
'C 'C
u. ~: "" s
Q <>
". 1 0
g" ~ "
Ii-
§ 0 o
<>. ~ ~ ~
~ l ~
~.
tt. :s
~. 1;1
!
g,.
" ~ f ~
~ 9
~(;:. ( 3
m
~. '"
'"
.,.,,0 ~ .
11 s'§
~ !?"O
o ]Jl. ::l
0
; 9. a ::l
""g ~ ~
zr
If~. (1)
}>
a: g :c
l! \J
i< • "w

•

II

,~ j § ~ 13
;i 0 ~
.§ 0
t!
:1Jil> ~ <l>
~g
~ B
Jr '" s
~~ S s
~_g !5
~ ~
,!, "
0 i3 :.a
~ :~ 'B ii
1:1
e "'" 3. 1
"
III 'B .§
"Cl "
(5 " " ,D
.c 'f ,.g !5
~ ... ~ ~
" Q
~ ~ " i' ~
"
" ~
.... " 1! " i
0 i3
+-' 0 ~
.c il &. 'g '&;
C) ,§ ~ 'g
~ "" -B
!! " <,.., "'" ,eo
.g,~ " 0 i 0
0 6 '~
'¥ ~ 8 '~ 0.
't1_ -< -<
~1 ~ 0
~ ..; ni
Ii if's/
l r-'-I.i~>~ ~
ill i'. ~~
,Pl ·~"hi·
1. H!H~~H~J.
. j-~{--~.;f<-.-';';. ~ " " ~!'l * '"
o c:
c: > III > oJ)
<II E ~]
0 :::>
~ 0 <II III
c: c:Ol ~.t:
'" ~li .,0.
~ .., c:
'" c: c: t: 0
Ol 80 0
ro 0
'l'l ~.2l c:
1a IJJ c: m <J)
E E IJJ -g E ,!l!
IJJ '" o E 1:.2 ~
s: ~ 'c OJ
1Q - '" .Ql a:; 'C
IJJ-o ~ E <>-
~ 5'~ ,s c:
,_ 0
(llo ~
'"
c '"
oJ) IJJ
E ~
III ,Q
Ol
'" li
,2-
Aggregation on •

- . .

...... ......

;;: ~ •
6' " 0 ..
5- •
!1 <if c
a: " a % Ii
8 " ;;;f
"
S "
"
a
~ 13 ~ c,
<J; a
0
" ~
0
"
~ '" " '"
<J; <J; !¥ 3
11
;;: 0
!11. ....,
" '0
~ ~ vr '" <if ~:
"
§
g C>
'"'
0
~ ~ '" '" .g
!t
g
<> <> ~ <> '"
~ ~ !:l il
.. "' .. ... ll.
i

§
.,
-!3
d
0
.,
'(;l
;>,
Ol
§
·f
. .,
.~
'"
<I}
U
•
....
• ~
~ t! 0"> !:i:
"<l:
cr d M 0
.,
~
~ r-. (0 r-,
r-. (J) ('.j
e ('.j "<l: N
~ d 0 d
t1 ~ ;t <0
CO
0 M v- "<l:
o d d 0
Q
Q 0
~ '<=1
.~·s a .g ~
""
gJ §~ ]i§ ]i]
</)C\i ;:;:;~ ;:;:;~ ;:;:; u

•

....

•

•
.....
•
~ II
~ , '"
, , ~ f
~'< .. / ~ "- -, ,
"- <, H
/ , ~
, ~
, ~ f~ ............ __ t
J ,,~ ~
~
~ >
l
e:
@.
" ~ ~ [ .@
i • i
~ ~
:;- :;- ""0 ~.
o Q. ~
0 ~
3 "
'0 @ OJ
to :::J
til :::J o
g £ "
~ s •
.....
•
p Ii
~
'"
-, n ~~
<,
-, " ~ ~~
~ '" "'g i
"'- ... f-~
......
~ ~'" e:
<>
~.[ e.
1
~ ct.
--_ >- s
-_ 1·[ 0
; '"
0
Ei
0
[:l
<>
0:1 a
o g ~
~ ,., ,.. '1':,::~t~ ~ !.".
f ~'';'1:'''lffif''
~ ~ ~_~ _ .,:i:.:- __ ft': •
">" ~ ~ .• ,....8 .•. § '" •. ' . n
' ". '''f~:ll''''''
······.~IF.~f) •• ··•··.
~~
f- ~ t~ ,...
3 W s 2 !l' a 8 r 11 '"
~ :;.e ;T a ~ ~:
R ~ ~. s 9 ~ " ~
a 0 0 " !to
~ ~ ~ ~ 1;1 C>. a
~
~ 3o!! ~ ;! r if s ~
e.g. 0 r g- s '"
= .., " a
~ oil- :;. ~ g I
~ P.-g o o· g
'" ~ g- ~~ "
c C>. fl. s £, "
... ~ ~ '" [
'" a~ ;. ! ~ <to
'" Ii
!l ;! c. e
f.':: !l a ,., o·
g. ;! ;;! ~ § fl
~. R g'~
" Q b is Q.
e, t.l' :::0 ~ a 8. "
119 l ~
g- ~ ~ 0 0'
6'
~g. ~ ~
"- ~. " P.- '" s:
fl. g 0" s· ~ s-
!!. "
0 [g- OJ "
I:; p C>.
u E
&: s:
" AHPSort for selection and evaluation of a large number of suppliers

Alessio Ishizaka, Craig Pearman

UniNlJily r1FoltJIllOlllh, FOJ7J!JJolllh BIISilleJS Schoo/, RichJJJond Bllildi"g, PO 1 3ED Pot1J1JJO!ff/;, United KiNgdolll Vllmio. hhii(,[email protected]:llk; CraigPeatlJlflll@botllJailfO./lk)

Abstract: The Analytic Hierarchy Process (AHP) is a useful method but it is not often applied for supplier selection, especially when the number of candidates is high. This p;lper presents a new two stage model, namely the AHPSort and the ,\HP. First, the AlH'Sort;« new variant of the AHP, is used for sorting alternatives. Qualified suppliers are then ranked with AEP. A real case stud>, is used to evaluate our approach. However useful, two conditions are missing for a full adoption by the studied company: a full explanation of the mathematics behind the method and the implementation in an integrated procurement system.

Keywords: AHP, AHPSort, J\1CDA; supplier selection, outsourcing

1. Introduction

Suppliers are of tremendous importance to their clients. This importance is accentuated with the pressure to reduce the supply base (Ogden 2006; Sarkar and Mohapatra 2006). Wrong supplier selection could lead to serious consequences. Hoecht and Trott (2006) have enumerated several problems due to poor supplier and unhealthy buyer-supplier relationships. Recent surveys confirm the importance given to the supplier selection process (Lieb and Bentz 2005; Lieb and Bentz 2006; Lieb and Butner 2007). Therefore, several techniques have been developed to support this exercise (Aissaoui ct al. 2007; de Boer et aL 2001; El-Sawalhi et al, 2007). However the Analytic Hierarchy Process (AHP) has it high number of success stories in many areas, sec compilations of (Forman and G;lSS 2001; Golden et al, 1989; Ho 2008; Kumar and Vaidya 2006; Liberatore and Nydick 2008; Omkarprasad and Sushil 2006; Saaty and Forman 1992; Shim 1989; Sipahi and Timor 2010; Vargas 1990; Zahedi 1986), however it is rarely used when the number of suppliers arc largc. TIle main reason for this is the litrgc number of pairwise comparisons required when the number of alternatives is high. In this paper, we will present a method to decrease the number of pairwise comparisons required. In this paper we will first summarise the benefits of AHP and then introduce AHPSmt for sorting a large number of suppliers. It W;lS found that AHPSort and AHP, if used successively, can drastically decrease the number of pairwise comparisons, without deteriorating the advantages of AHP.

2. AHP

The first step of AHP is to structure the problem. As with any other Multi-Criteria Decision Aid (i'.ICDA) method, the problem is divided into the goai, alternatives to achieve this goal and criteria to evaluate the performances

of the alternatives. The particularity of AHP is that criteria can be divided into sub-criteria, which forms a hierarchy, This subdivision allows the decision-maker to express his judgements on fewer criteria at anyone time, which should give more precise evaluations.

'111en, each alternative is evaluated pairwise in a comparison matrix. This again allows the decision-maker to concentrate only on two alternatives at that time. Local priorities are then calculated from these matrices with the eigenvalue method:

A -p= x vp

(1)

A is the comparison matrix P is the priorities vector

). is the maximal eigenvalue

In order to declare the comparison matrix consistent enough for calculating credible priorities, it must pass a consistency check. If the Consistency Ratio (CR) is less than 10%, then the matrix can be considered as having an acceptable consistency.

where

CR:=: CI/RI

(2)

where RI is the Random Index (the average CI of 500 randomly filled matrices)

CI is the Consistency Index

CI = (Ami< - /J)/(11-1)

(3)

where JI :=: dimension of the matrix

A".". :=: maximal eigenvalue

Other derivation methods exist, however simulations have shown that differences in the calculated priorities are 110t significant (Budescu, Zwick, & Rapoport, 1986; Cho & \'\!edley, 2004; Golany & Kress, 1993; Herman &

Koczkodaj, 1996; Ishizaka & Lusti, 2006; Jones & Mardle, 2004; Mikhailov & Singh, 1999).

The same process of pairwise comparison is applied for each group of sub-criteria and criteria in order to calculate the weight of the criteria. Finally, the global priority of each alternative is calculated by a weighted sum of the local priorities . More details on the method can be found in (Saaty, 1977, 1980; Ishizaka and Labib, 2009).

3. AHPSort

As AHP is not adapted for a large number of suppliers due to its high number of pairwise comparisons required, we first use AHPSmt in a screening stage (figure 1).

limiting profile for c!ass1

~~~Stf ~~alified SUPPliers)

Class 3: j

Set of unsucessful ••

suppliers

limiting profile for class 2

Figure 1. AI-IPSo!t as a screening stage

The AI-IPSort method is based on eight steps:

1) Define the criteria to assess each candidate.

2) Evaluate pairwise the importance of the criteria and derive weights with the eigenvalue method as in AHP (see section 2).

3) Define the classes used to sort the candidates.

4) Define the limiting profile for each class: what is the minimum performance needed on each criterion to belong to a class.

5) Compare in a pairwise comparison matrix a single candidate with the limiting profiles for each criterion.

6) Derive local priorities with the eigenvalue method as in AHP for each criterion.

7) Aggregate the local weighted priorities, which provide a ranking used to allocate a candidate to a class.

8) Repeat process 5) to 7) with the next candidate.

4. Case study 4.1 IlJlrodlfdioJl

The studied organisation is required by Royal Charter to formally report on its activities and performance in relation to income and expenditure. The organisation itself provides a diverse range of services and is supported by a number of wholly owned subsidiaries, each operating

on a full commercial basis and providing services to the organisation, making reporting a complex process. Any report also has to incorporate multiple sign off stages and needs to be flexible enough to accommodate a full Welsh version, which can only be produced once the English version is finalised, yet is required to launch at the same time. The choice of a supplier to produce these publications is the specific organisational problem that will be analysed.

4.2 Pre-qlltl/ift(alioJl slage

In the pre-qualification stage, the company selects all suppliers which fulfil the minimal requirements for the vacancy, in order to progress to the invitation to tender stage. Five criteria have been selected by the management:

Experience: The organisation requires an established company for this contract.

Flexibility: The selected company should be able to offer a large range of services and to be contactable at an>' time during the working week.

Security: The purpose of this criterion is to ensure project control on quality, timings and physical security considerations.

Resilience: This criterion relates to the suppliers realistic potential to complete the project, without risk or flaw.

Environment: This criterion considers any environmental accreditations the tendering organisation may have.

The criteria are pairwised compared (table 1) and priorities are derived (table 2).

Exp Flex Sec Res Env
Experience 1 7 4 7 8
Flexibility 1/7 1 1/4 1 3
Security 1/4 4 1 4 5
Resilience 1/7 1 1/4 1 3
Environment 1/8 1/3 1/5 1/3 1 Table 1. Comparison matrix of the criteria in the prequalification stage

Criterion Weighting
Experience 0.565
Flexibility 0.081
Security 0.234
Resilience 0.081
Environment 0.040
L:i~J i .•• KltliOYii I •• ·"· •. • ... ••••·· •• •• ... ······.····{) , ..... y ......................... Table 2. \,\leights of the criteria

111en suppliers arc evaluated in regards to each criterion, against the limiting profile, Their rankings arc given in table 3. Suppliers in grey are below the limiting profile and therefore do not progress in the next stage.

Score AHPSort

.900

.900

.894

.809

.791

Table 3. Supplier ranking in the qualification stage 4.3 Evalllatioll stage

This stage aims to identify the most suitable supplier to offer the tender to, from those who progressed. Five criteria have been selected by management for this stage:

Quality: The tendering companies will be evaluated with respect to their skills base. Skills include, but arc not limited to, ongoing staff training, company focus, creative thinking, strategic thinking, complex stakeholder management, clear understanding of the challenge, success and risk factors, company culture, and team fit.

Value for Motley: This criterion relates to the services offered relative to the price quoted, and is the organisations opportunity to analyse the price of the tender along with other financial implications.

Contract Management: The organisation requires the tendering company to show its ability to manage relationships with suppliers, its handling of sensitive information, print management, contract implementation, methodology for service and contract delivery, quality of methodology and project planning processes, quality of structure of programme to deliver on time and under budget.

Client Service: The tendering company is expected to demonstrate the willingness, ability and resource to work flexibly and to respond to tight deadlines.

Environment: This criterion relates to the suppliers demonstration and evidence of physical office security, physical plant security, document storage, pl'Ocess for dealing with online and 'soft' copy information and project resilience.

The criteria are pairwised compared (table 4) and priorities are derived (table 5).

Qual VfM Ci\f CS Env
Quality 1 5 7 7 9
Value for 1/5 1 3 3 5
money
Contract 1/7 1/3 1 1 3
management
Client service 1/7 1/3 1 1 3
Environment 1/9 1/5 1/3 1/3 1 Table 4. Comparison matrix of tile criteria in the evaluation stage

Criterion

AHP weighting

Qua~_!}~

0.598

Value for mons:

0.195

Contract management

0.084

Client service

0.084

Environment

• •• Inconsisteh9'.RiH1oi>

0.039

Table 5. Criteria weighting

Suppliers are evaluated on the above criteria using the AHP. All inconsistency ratios are below 0.1. The final result set is described in table 6.

Supplier ScoreAHP
SuQ1>1ier C .258
S~lierA .258
Supplier B .148
Supplier E .110
S~crD .145
Supplier F .065
Supplier G .017 Table 6. Final ranking of the suppliers

Supplier C and A have the same final score. Finally, in an additional negotiation stage, supplier A was selected because it offered a lower tender cost.

5. Conclusions

Decisions in business arc of significant importance. This study has used the supplier selection exercise as all example of a contemporary decision, which warrants an extensive, transparent and critical process.

In the literature review, we have seen that the AHP is a widely used method but is rarely applied for supplier selection with a high number of candidates due to the high number of pairwise comparisons required. In order to tackle this problem a two stage model incorporating AHPSod and the AHP has been developed and tested using a real case. \,\/hen observed in comparison to the process already in operation at the organisation, the AHP was more accurate in the preference elicitation and was not time consuming,

After discussion with the case study organisation in the post-analysis, it was dear that management were impressed with certain asp ects of the model, especially the higher accuracy in the preference elicitation. However

they communicated two primary concerns that would prevent them from incorporating it into their own process. The first concern was the difficulty to use this model as it is not integrated into their current information system, which also collects and stores data, records performance of the suppliers, manages documents, communicates with the supplier, sends offers, produces reports, etc. The second concern was regarding the effect that the EU imposed regulation would have on the model, specifically the requirement for a consistent process.

The first concern is understandable. Providing decisionsmakers with user-friendly software is essential for the adaptation of a method. The second concern is much easier to redress. In fact we did not teach the mathematical background of the AHP to the company, in particular they are not aware of an inbuilt consistency check. It is totally legitimate that a company has some reticence to adopt a method viewed as a black box. The provision of trainings on the AHP and the AHPSort is therefore necessary.

Our study has opened the gate to a wider use of the AHP in supplier selection by reducing the number of pairwise comparisons needed through the use of a two stage approach, namely AHPSort-AHP. Two other gates still need to be opened: the enhancement of its credibility with trainings and its practicality by an implementation in an integrated procurement system.

References

Aissaoui, N., Haouari xr, and Hassini E. (2007). Supplier selection and order lot sizing modeling: A review. COIJIjJJllerJ & OperatioJIJ Rtsfmrh, 34(12), 3516-3540.

Budescu, D., Zwick, R., and Rapoport, A. (1986). A comparison of the eigenvalue method and the geometric mean procedure for ratio scaling. App/ied p!),chologiral JfledsJlm!Ulll, 10(1),69-78.

Cho, E. and Wedley, W. (2004). A Common Framework for Deriving Preference Values from Pairwise Comparison Matrices. Comp/ders and Operatio»: Remllrh, 31(6),893-908.

De Boer, L., Labro, E., and Morlacchi, P. (2001). A review of methods supporting supplier selection. European j01l17/a/ 0/ Plllrhasillg & SIPPI)' j\JaJlagOlIlNlt, 7(2), 75-89.

El-Sawalhi, N., Eaton, D., and Rustorn, R. (2007). Contractor pre-qualification model: State-of-the-art. International jOllmai 0/ Project MallagtlJleJIf, 25(5), 465-474.

Forman, E, and Gass, S. (2001). The Analytic Hierarchy Process - An Exposition. Operations Research, 49(4), 469- 486.

Golany, B., and Kress, M, (1993). A multicriteria evaluation of the methods for obtaining weights from ratio-scale matrices. Ellropean [ouma! <ifOperaliofial Reseanb, 69(2), 210--220.

Golden, n, Wasil, E., and Harker, P. (1989). The Analytic Hierarchy Process: applications and studies. SpringerVerlag, Heidelberg.

Herman, M. and Koczkodaj, W. (1996). A Monte Carlo Study of Pairwise Comparison. bt/omlalioll Processing Letters, 57(1),25-29.

Ho W. (2008). Integrated analytic hierarchy process and its applications - A literature review. European [onrna! III Operatiollai Restaffh, 186(1), 211-228.

Hoecht, A. and Trott, P. (2006). Outsourcing, information leakage and the risk of losing technology based competencies. EIII'OPMll BIIJbltSS Review, 18(5),395-412.

Ishizaka, A. and Lusti, 1\L (2006). How to derive priorities in AHP: a comparative study. Central BIIJ'/)peaJl [ouma! 0/ Operations Rman'h, 14(4), 387-400.

Ishizaka, A. and Labib, A. (2009). Analytic Hierarchy Process and Expert Choice: benefits and limitations. OR IIiSight, 22(4), 201-220.

Jones, D. and Mardle, S. (2004). A Distance-Metric Methodology for the Derivation of Weights from a Pairwise Comparison Matrix, JOlin/II! <if the OpemfioJlal Research Sotiety, 55(8), 869-875.

Kumar, S. and Vaidya, 0. (2006). Analytic hierarchy process: An overview of applications. EIIJ'ojJMI/ [onmal q/ Opemtjol/(/IRtsemrh, 169(1), 1-29.

Liberatore, M. and Nydick, R. (2008). The analytic hierarchy process in medical and health care decision making: A literature review. Em'ojJMII [ourua! q( Operational Researeb, 189(1),194-207.

Lieb, R. and Bentz, B. (2005). The North Amarican third party logistics in 2004: the provider CEO perspective. International [osrna! of Pbysk»! Distribution & ugiJIiCJ j\Jallagt!!!Jellt, 35(8), 595-611.

Lieb, R. and Bentz, B. (2006). The 3PL Industry in Asia/Pacific. SlpP!)' Chain MallageJJlmt Rt!,jeJl', 10(9), 10-15.

Lieb, R. and Butner, K. (2007). The North American Third-Party Logistics Industry in 2006: The Provider CEO Perspective. Transportation [oumal; 46(3), 40-52.

Mikhailov, L. and Singh, i\L G. (1999). Comparison analysis of methods for deriving priorities in the analytic hierarchy process. IEEE Internatlona! C01ifmllct 011 Systems, Mall, and (),bemetits, 1, 1037-1042.

Ogden, J. (2006). Supply base reduction: an empirical study of critical success factors. [ourna! ~( SlIppb' Cbain MalJagcJtlml, 42(4), 29-39.

Omkarprasad, V. and Sushi!, K. (2006). Analytic hierarchy process: an overview of applications. Enropenn [ouma! 0/ Operatlona! Rmmrb, 169(1), 1-29.

Saaty, T. (1977). A scaling method for priorities in hierarchical structures. j onmal 0/ Mathematical PS)'ch%gy, 15(3), 234-281.

Saaty, T. (1980). '[heAl/a!ylie Uieralli!)' PrOCliSJ. McGrawI-fill, New York.

Saaty, T. and Forman, E. (1992). The Hierarrbon: A Die/lolld!)' tf Hieranbies. RWS Publications, Pittsburgh.

Sarkar, A. and Mohapatra, P. (2006). Evaluation of supplier capability and performance: A method for supply base reduction. [ouma! of PlIIrvasillg fwd Sftppfy iHaJldgtJJMJ//, 12(3),148-163.

Shim, J. (1989). Bibliography research on the analytic hierarchy process (AHP). Sotio-ECOtJOJ)fit Pklll1lilJ,g SdtJlM, 23(3),161-167.

Sipahi, S. and Timor, M, (2010). The analytic hierarchy process and analytic network process: an overview of applications. lIIatlageJJleJIf Dedsios, 48(5), advande online publication.

Vargas, L. (1990). An overview of the analytic hierarchy process and its applications. EmvpetllJ [ourna! ~f Operatlona! Reseanb, 48(1), 2-8.

Zahedi, F. (1986). The analytic hierarchy process: a survey of the method and its applications. IJlteJjtue, 16(4), 96-108.

Selection of new production facilities with the Group Analytic Hierarchy Process Ordering Method

Alessio Ishizaka*, Ashraf Labib

University of Portsmouth, Portsmouth Business School, Richmond Building, Portland Street, Portsmouth POI 3DE, United Kingdom

EmaiIs:[email protected];[email protected]

Abstract:

This paper presents the Group Analytic Hierarchy Process Ordering (GAHPO) method: a new multi-criteria decision aid (MCDA) method for ordering alternatives in a group decision. The backbone of the method is the Analytic Hierarchy Process (AHP) which is separated into two hierarchies for a cost and a benefit analysis. From these two analyses, a partial ordinal ranking can be deduced, where three relations between alternatives exist: the preference, indifference, and incomparability. A complete cardinal ranking can also be deduced by dividing the score of the benefit analysis by the score of the cost analysis. Another particularity of GAHPO is the incorporation of 'fairness' when assigning weights to the decision makers. GAHPO has been developed to solve a real case: a selection of new production facilities with multiple stakeholders. By applying this method, we found four main advantages: significant reduction oftime and effort in the decision process; easiness for the decision makers to arrive at a consensus; enhancement of the decision quality and documentation with justification of the decision made.

Keywords: Group Analytic Hierarchy Process Ordering (GAHPO), Group decision, Cost/benefit analysis, Incomparability, Facilities selection

1. Introduction

Strategic decisions are fundamental to any company. They are usually not determined by a single decision-maker but by a group of decision-makers, who may have different objectives. In this case, two distinct methodologies are commonly used (Srdjevic, 2007): multicriteria decision-making methods or voting system. The voting system has surely high democratic properties and bypasses the data requirements of multicriteria approaches (Hurley & Lior, 2002) but moves stakeholder into a polarisation of their opinion and no intensity of their preferences can be measured. It is a head-count of yes or no. Therefore, a minority with strong convictions will unconditionally be beaten from a majority, whatever the strength of their opinion is. Furthermore, a voting system does not necessitate a modelling of the problem and therefore has difficulty to incorporate several criteria in the decision (Craven, 1992). Saaty and Shang (2007) recommend using AHP in order to resolve deficiencies of the conventional voting mechanism. AHP is a multi-criteria method developed by Saaty (1977; 1980) and applied in several area: banks (Secme, Bayrakdaroglu, & Kahraman, 2009), manufacturing systems (Ic & Yurdakul, 2009; T.-S. Li & Huang, 2009; Yang, Chuang, & Huang, 2009), operators evaluation (Sen & CInar, 2009), drugs selection (Vidal, Sahin, Martelli, Berhoune, & Bonan, 2009), site selection (Onut, Efendigil, & Soner Kara, 2009), software evaluation (Cebeci, 2009; Chang, Wu, & Lin, 2009), evaluation of website

* corresponding author

Acknowledgements: We are grateful to Binoy Perumpalath, who helped to collect the data for the case study.

performance (Liu & Chen, 2009), strategy selection (Chen & Wang, 2009; S. Li & Li, 2009; Limam Mansar, Reijers, & Ounnar, 2009; Wu, Lin, & Lin, 2009), supplier selection (Chamodrakas, Batis, & Martakos; H. S. Wang, Che, & Wu; T.-Y. Wang & Yang, 2009), selection of recycling technology (Y.-L. Hsu, Lee, & Kreng), firms competence evaluation (Amiri, Zandieh, Soltani, & Vahdani, 2009), weapon selection (Dagdeviren, Yavuz, & KiIIn9, 2009), underground mining method selection (Naghadehi, Mikaeil, & Ataei, 2009), software design (S. H. Hsu, Kao, & Wu, 2009), organisational performance evaluation (Tseng & Lee, 2009), staff recruitment (Celik, Kandakoglu, & Er, 2009; Khosla, Goonesekera, & Chu, 2009), construction method selection (Pan, 2009), warehouse selection (Ho & Emrouznejad, 2009), technology evaluation (Lai & Tsai, 2009), route planning (Niaraki & Kim, 2009) and many others. This paper presents the Group Analytic Hierarchy Process Ordering (GAHPO), which improves the AHP on several points. We separate the cost and benefit criteria of the AHP, which simplify the appraisal and provide a more accurate result, as will be shown later. Results are then partially aggregated for an ordinal partial ranking or fully aggregated for a cardinal complete ranking. The new GAHPO method is also adapted for group decisions. The task to assign weights (importance) to the different decision-makers of the group is often a difficult one. We propose a new simple and fair method, where the weights of the members are judged by the other members of the group.

The paper starts with a literature review on the Analytic Hierarchy Process, followed by the description of the new proposed method and then finalised by an application of production facilities selection.

2

2. Analytic Hierarchy Process

AHP decomposes the problem into small parts in order to facilitate the decision-maker in the appraisal task. First, a hierarchy structuring the problem is constructed (figure 1), The top of the hierarchy represents the goal. Below we have the criteria, sub-criteria and alternatives. The appraisal can be constructed top-down or bottom-up (figure 2) but always using pairwise comparisons. It allows the user to concentrate only on the question "How much A is better than B't" and to ignore temporary the other criteria and alternatives. The comparisons are entered into a matrix. If a matrix is sufficiently consistent, priorities can be calculated with the formula:

Aw = Ama.x:W (1)

where A comparison matrix

"ma.x: principal eigenvalue

w vector of the priorities

Figure 1: Hierarchy used in the AHP.

TOP DOHW + /JOn'OM UP

MORE GENEBAL

I

• • •

• •

•

3

d If Z'lfRiVd VVRS

MORE SP£!;JHC

Figure 2: Top down or bottom up appraisal (Chan & Chan, 2004).

The comparison matrix contains redundant information. This redundancy serves the purpose of refining the final result as it makes the approach less dependent on one single judgement. The AHP model provides a feedback to the decision maker on the consistency of the entered judgements by a measure called consistency ratio (CR):

CR= C] RI

(2)

A. -11

and C]= max

n-I

where CI consistency index

11 dimension ofthe comparison matrix

Amax principal eigenvalue

RI ratio index

(3)

The ratio index (RI) is the average of the consistency index of 500 randomly generated matrices. If the consistency ratio is higher than 10%, it is recommended to revise the

comparisons in order to reduce the inconsistency. Once all local priorities are available, they are aggregated with a weighted sum in order to obtain the global priorities of the alternatives.

3. Analytic Hierarchy Process Ordering

Later, it was proposed (Azis, 1990; Clayton, Wright, & Sarver; Wedley, Choo, & Schoner, 2001) to decompose the model into further subproblems, in separating criteria with opposite direction in different hierarchies: benefits versus costs. The reason of this additional decomposition is that criteria on the same direction are much easier to compare than two in opposite directions like a criterion to be minimised and another to be maximised. In this paper, we introduce the concepts of partial ordinal ranking and complete cardinal ranking.

In some problems, an order of alternatives is sufficient to take a decision. A partial ordinal ranking can be derived from the cost and benefit analysis, where:

1. Alternative A is better than Alternative B if Alternative A is ranked better than Alternative B in the cost and benefit analysis.

4

Alternative B Alternative A
0
/ I
/ /
/ /
/ /
/
/ / Benefit

Cost

o

Figure 3: Graphical representation of the preference relation

2. Alternative A is indifferent to Alternative B if Alternative A has the same score than Alternative B in the cost and benefit analysis.

Alternative A Alternative B

Benefit

o

/ / /

/

Cost

o

Figure 4: Graphical representation of the indifference relation

3. Alternative A is incomparable to Alternative B if Alternative A is better in one analysis and worst in the other analysis.

Alternative B

Alternative A

Benefit

o

Cost

o

Figure 5: Graphical representation of the incomparability

Incomparability does not exist in the standard AHP. This status is important as it reveals that we cannot decide which of two alternatives is the dominant one: an alternative is better on some aspects but worst on others. In order to decide, which alternative is better, further discussion between the decision-makers moderated by the analyst is needed. This further debate may require additional information. However if a debate cannot be hold, for example because the decision-makers are unavailable, the cost and benefit analysis can be merged in one ranking by dividing the score of the benefit analysis by the score ofthe cost analysis. This produces the complete cardinal ranking.

Alessio, we are assuming here that benefits and costs are equally important. We are also assuming that benefits incorporate opportunities and costs incorporate risk. If that is correct, should we mention this somewhere in the text?

4. Group decision

As a decision affects often several persons, the standard AHP has been adapted in order to be applied in group decisions. Consulting several experts avoids also bias that may be present when the judgements are considered from a single expert. There are four ways to combine the preferences into a consensus rating (table 1).

5

mathematical aggregation
Yes No
s:I judgements geometric mean on judgements consensus vote on judgements
0
'+3
i:I::I ••
bO s:I
~o priorities weighted arithmetic mean on consensus vote on priorities
bO priorities
ro Table 1: Four ways to combine preferences.

The consensus vote is used, when we have a synergistic group and not a collection of individuals. In this case, the hierarchy of the problem must be the same for all decisionmakers. On the judgements level, this method requires the group to reach an agreement on the value of each entry in a matrix of pairwise comparisons. A consistent agreement is usually difficult to obtain with increasing difficulty with the number of comparison matrices and related discussions. In order to bypass this difficulty, the consensus vote can be postponed after the calculation of the priorities of each participant. O'Learly (1993) recommends this version because an early aggregation could result "in a meaningless average

performance measure". An aggregation after the calculation of priorities allows to detect decision-makers from different boards and to discuss further any disagreement.

If a consensus is difficult to achieve (e.g. with a large number of persons 01' distant persons), a mathematical aggregation can be adopted. Two synthesizing methods exist and provide the same results in case of perfect consistency of the pairwise matrices (T. L. Saaty & Vargas, 2005). In the first method, the geometric mean of individual evaluations are used as elements in the pairwise matrices and then priorities are computed. The geometric mean method (GMM) must be adopted instead of the arithmetical mean in order to preserve the reciprocal property (Aczel & Saaty, 1983). For example, if person A enters a comparison 9 and person

B enters 119, then by intuition the mathematical consensus should be ~9 ·i =1, which is a geometric mean and not (9 + 119)/2 = 4.56, which is an arithmetic mean. Ramanathan and Ganesh (1994) give an example where the Pareto optimality (i.e, if all group members prefer A to B, then the group decision should prefer A) is not satisfied with the GMM. Van den Honert and Lootsma (1997) argue that this violation could be expected because the pairwise assessments are a compromise of all the group members' assessments and therefore it is a compromise that does not represent any opinion of the group member. Madu and Kuei (1995) and then Saaty and Vargas (2007) introduce a measure of the dispersion of the judgements in order to avoid this problem. If the group is not homogenous, further discussions are required to reach a consensus.

6

In the second method, decision-makers constitute the first level below the goal of the AHP hierarchy. Priorities are computed and then aggregated using the weighted arithmetic mean method (WAMM). Applications can be found in (Labib & Shah, 2001; Labib, Williams, & O'Connor, 1996). Arbel and Orgler (1990) have introduced a further level above the stakeholders' level representing the several economics scenarios. This extra level determines the priorities (weights) of the stakeholders.

In a compromised method individual's derived priorities can be aggregated at each node. However according to Forman and Peniwati (1998), this method is "less meaningful and not commonly used". Aggregation methods with linear programming (Mikhailov, 2004) and Bayesian approach (Altuzarra, Moreno-Jimenez, & Salvador, 2007) have been proposed in order to take a decision even when comparisons are missing, for example when a stakeholder does not feel to have the expertise to judge a particular comparison.

Group decision may be skewed because of collusion or distortion of the judgements in order to advantage its preferred outcome. As individual identities are lost with an aggregation, we prefer to avoid an early aggregation. Condon, Golden, & Wasil (2003) have developed a programme in order to visualise the decision of each participant, which facilitate the detection of outliers.

5. Weight of stakeholders in GAHPO

If all decision-makers do not have an equal weight, their priority must be determined. The weights reflect the expertise of a decision-maker (Weiss & Rao, 1987) or the importance of the impact ofthe decision on the decision-maker. The weights can be allocated by a supra decision-maker or by a participatory approach. Finding a supra decision-maker or benevolent dictator, which is accepted by everybody, may be difficult. Cho and Cho (2008) have a surprising way to determine the weights with the level of inconsistency. We do not support this method because the inconsistency is a useful feed-back to the user. It indicates to the decision maker his/her consistency, recommend revision of comparisons that maybe due to a

manual error in setting the comparisons, sometimes forced due to the upper limitation of the comparison scale (e.g. if the user enters first a12 = 4 and a23 = 5, he should enter a]3=20 in order to be consistent, but he can only enter a]3=9 due to the maximal value of the measurement scale). The consistency index is therefore certainly not a measure of the quality or expertise of the decision-maker. Ramanathan and Ganesh (1994) have proposed a method based on pairwise comparisons to calculate the weights. All n members fill a comparison matrix with their relative importance of each participant. A vector of priorities is calculated for each member. The n vectors of priorities are gathered in a n x n matrix and the final weight of each member is given by the eigenvector of this matrix. In order to incorporate the uncertainty of the expertise of the participants, the AHP has been combined with variable precision rough set (Xie, Zhang, Lai, & Yu, 2008) and fuzzy logic (Jaganathan, Erinjeri, & Ker, 2007).

The GAHPO also uses pairwise comparisons but only to judge other members of the group, with a veto possibility by the evaluated persons. This technique can be viewed as more fair and is applied for example in ice skating, where judges cannot evaluate competitors of the same nationality. The consistency of the weights given by the appraisers is checked with the consistency ratio formula (3).

6. Methodological approach

The case study took place in a world leading packing company, which had no previous experience in multicriteria methods. Our approach was based on four phases, each one corresponding to a meeting with the decision-makers of the company, where the researchers where facilitating the decision process.

a) An awareness session on the GAHPO methodology was given. An understanding of the GAHPO and required inputs is necessary in order to avoid improper use of the method (Cheng, Li, & Ho, 2002). The advantages of the new decision method were clearly explained in order that everybody accepts it and to avoid reluctance and objections during the decision process.

b) After a brief reminder on the GAHPO, the problem and its possible solutions were clearly defined. Two hierarchies were constructed: one for costs and another for benefits.

c) At the beginning of the third meeting, the participants were given the opportunity to revise the hierarchies. Then, each participant gave its comparisons of alternatives, criteria and participants' weights through a questionnaire. The participants' weights were given by the other group's members. Consistency was checked for each participant.

d) Priorities are aggregated in Expert Choice. A sensitivity analysis is conducted and results are discussed.

7

7. A case study: selection of new production facilities

The studied packing company has two plants in England: the 'Green' plant producing paper products and the 'Plasto' plant producing plastic items. Due to a repatriation of another production plant from Scotland, the Plasto plant has to be redesigned. Three alternatives are possible:

1) Redesign of Plasto plant, hereafter referred to as Plant Redesign

2) Automation of Plasto production processes, hereafter referred to as Plant Automation

3) Relocation and consolidation of Green plant with Plasto, hereafter referred to as Plant Consolidation

We will now comment on the four phases of our methodological approach as described above.

7.1 An Awareness Session on the GAHPO

An half day awareness session was given to all stakeholders involved in the decision process (see next section for the list). The methodology of GARPO without the mathematics (too complicated for the audience) and an example with Expert Choice was presented. The advantages of the GARPO were clearly perceived. This first step is fundamental because, the way a new method is presented (and then used) can significantly impact its efficacy. The investment in time and money of using GARPO and its supporting software was approved due to the strategic importance of the decision, It was decided to continue with the next phase.

7.2 Structure of tile hierarchy model

A logically constructed hierarchy is the backbone of the entire GARPO approach, which means the GARPO is both a problem solving and a problem-structuring tool. The cost analysis and benefit analysis hierarchies were developed as two separate ARP models (see figures 6 and 7) in a half day brainstorming session with all stakeholders facilitated by the researchers. The cost analysis model has the goal of selecting the alternative with the lowest cost. The benefit analysis model has the goal of selecting the alternative with the highest benefit. The validity of the hierarchies was assessed by asking whether the elements of an upper level can be used as common attributes to compare the elements in the level below.

The first or uppermost level identifies the stakeholders: Shareholders, Senior Managers and Middle Managers - those who have the most influence and involvement in the organisation's decision-making process. The second level is a subdivision of the decision makers in middle management of the first level. The third level is concerned with the main criteria or objectives that affect the new production facilities selection and the last level shows the three strategic alternatives. Both cost and benefit hierarchies share the same elements in all levels except the third one. The elements of the various levels are explained in detail below:

8

Figure 6: Cost analysis hierarchy.

Figure 7: Benefit analysis hierarchy.

9

Stakeholders (Levelland 2)

The identification of the stakeholders (actors) was straightforward in this case. All the three alternatives called for high financial investment. This required the approval of the Shareholders. The Senior Managers make the strategic decisions for the company and obtain the funds required to implement those plans. The Middle Managers implement the strategic plans as well as help the Senior Managers in the planning process. Four Middle Managers:

Production; Quality, Maintenance & Facilities are included in level two. Although they fall in the same level in the organisational hierarchy, each have dissimilar stakes; preferences and power in organisational matters. Prioritisation of their stakes was essential for a high-quality decision. This is achieved by adding a separate level in the hierarchy.

Criteria (Level 3)

This level shows the cost and benefit criteria. The logic of traditional cost-benefit analysis

was used in order to identify them. .

Cost criteria are those, which required direct 01' indirect spending from the company. Four kinds of costs were considered:

~ Capital as all the three alternatives would require significant capital investment, ~ Depreciation which depends on the type and duration of the investment,

~ Disruption costs are important as any alternative would cause an amount of disruption to the existing production process,

- Timeframe to implement the three alternatives were also considered as a key criterion because additional costs may be incurred due to cost inflation, currency risk etc. during the project life cycle.

The benefit criteria are those, which could attain quantitative or qualitative benefits to the company. Seven criteria were identified:

Work environment improvement could be beneficial to employee in their motivation and welfare at work,

Time saving through reduction in packing, palletising, and shipping time,

Labour saving can be gained by the reduction in the number of packers, operators and other workers,

Material saving is possible by reduction in start-up losses and scrap;

Quality improvement is achieved by a consistent production output and reduction in scrap, Company image can be gained with the top modern facilities,

Flexibility can be achieved through the ability to expand the production facility and sharing of resources.

10

Alternatives (Level 4)

Three alternatives are considered:

~ Plant redesign: it will allow accommodating new production lines.

- Plant automation: it will increase the capacity of production and safe costs in a long term.

- Plant consolidation: a total new plant incorporating the current Green and Plasto plants.

7.3 Assessment of pairwise comparisons

At the beginning of the third meeting, the participants were given the opportunity to review or revise the hierarchy model, but no modification was suggested. The next step collected the pairwise comparisons through written questionnaires. This method was selected in order that participants are not influenced by others' opinions. First, the stakeholders' weight was evaluated (table 2). The judgements were given by the other members of the group. For example, the Shareholders estimate the Senior Managers five times more important that the Middle Managers in this decision. If the evaluated participants feel that they are unfairly judged, they have a veto right and further discussion is then needed. In our case, there were a high consensus (CR=0.03 in table 2) and the veto right was not used.

The weights can be justified based on ultimate decision making power of stakeholders as follows. The Shareholders have higher importance compared to Senior Managers and Senior Managers have higher importance than Middle Managers do. Shareholders decide which strategy to invest the money, so they have the considerable influence. The Senior Managers decide which alternatives are to be proposed for investment, so they have strong importance, The Middle Managers decide whether they have a requirement for an alternative, so they have a low - medium importance.

Top level stakeholders (actors)
Stakeholder group Senior Middle Relative
Shareholders importance
Managers Managers
Shareholders 1 3 9 0.672
Senior Managers 1/3 1 5 0.265
Middle Managers 1/9 1/5 1 0.063
Consistency Ratio (CR) 0.03 Table 2: Pairwise comparisons of main stakeholders.

The relative importance ofthe members of the Middle Managers was assessed in a similar way (table 3). The comparison of two Middle Managers was given by the two other Middle Managers. There was low disagreement on the comparisons and a consensus was easily found between the appraisers. The Production Manager had the most influence in the company, while the Quality, Maintenance and Facilities Managers had lesser degrees of influence in a descending order.

II

Sub-group of Middle Managers
Sub-group of Relative
Middle Managers importance
Production Quality Maintenance Facilities
Production 1 3 5 5 0.538
Quality 1/3 1 5 5 0.305
Maintenance 1/5 1/5 1 1 0.078
Facilities 1/5 1/5 1/1 1 0.078
Consistency Ratio (CR) 0.06 12

Table 3: Pairwise comparisons of Middle Managers.

Then, each stakeholder evaluated the criteria. The process was straightforward with only few questions related to the supporting software. This suggests that our participants had no problem understanding and applying the pairwise comparisons technique. The priorities of the criteria from the stakeholders' point of view are presented in table 4 and 5.

Cost criteria
Stakeholder group Capital Depreciation Disruption Time-frame
Shareholders 0.447 0.053 0.105 0.396
Senior Managers 0.250 0.054 0.289 0.407
Middle Managers
Production Manager 0.175 0.060 0.383 0.383
Quality Manager 0.113 0.064 0.411 0.411
Maintenance Manager 0.083 0.083 0.417 0.417
Facilities Manager 0.113 0.064 0.411 0.411 Table 4: Assessment of cost criteria from the stakeholder point of view.

Benefit criteria
1:1 00 01)
il) .S ] 00 .~ '1:1 ;;.,. P
~ § g il)p [~ .....
Stakeholder group C<:l .......
tf.l ;.. ..... .....
il) .~ Bcd ,.c
o 0 «J g ~.§ '~
~ .~ il) j&
.S ~ ~ il)
~ 0 r--<
0 ~
<l) f-< H
Shareholders 0.310 0.111 0.044 0.262 0.071 0.033 0,169
Senior Managers 0.161 0.055 0,044 0.392 0,044 0.191 0.113
Middle Managers
Production Manager 0,052 0.231 0.231 0.231 0.091 0,054 0.111
Quality Manager 0.069 0,093 0.136 0.073 0,420 0,067 0,142
Maintenance Manager 0.137 0.137 0,137 0.119 0,199 0.104 0,166
Facilities Manager 0.254 0.073 0,073 0,064 0.064 0.236 0.236 Table 5: Assessment of benefit criteria from the stakeholder point of view.

For each criterion, the relative importance (local priority) with respect to the three strategic alternatives was entered from the viewpoint of each stakeholder, To illustrate this, the priorities of the Senior Manager are shown in table 6 and table 7,

Alternatives Cost criteria
Capital Depreciation Disruption Time-frame
Plant Redesign 0,202 0.143 0,618 0.258
Plant Automation 0,097 0.714 0.086 0.105
Plant Consolidation 0.701 0.143 0.297 0.637 13

Table 6: A Senior Manager's priorities for the alternatives.

Benefit criteria
1:1 00 00
,9 P
il) 0 ] 00 ~ '1:1 ;;.,.
,,..... il)p [~
~ § ~ .....
Alternatives r--<
il) ,9 til ;.. ....... .....
til BCil .o
o 0 ~ > !j S S .~
f$ .~ <l) j&
,g :2] ~ 0 o_ il)
~ 0 ........
E f-< ~
H
Plant Redesign 0.429 0.143 0.200 0,143 0.143 0.143 0.200
Plant Automation 0.429 0.714 0,600 0,714 0.714 0.714 0.200
Plant Consolidation 0.143 0.143 0,200 0.143 0.143 0,143 0.600 Table 7: A Senior Manager's priorities for the alternatives.

7.4 Calculation of priorities and sensitivity analysis

In the last step, Expert Choice was used for the calculation of the priorities and the sensitivity analysis. The participants had no difficulty in this phase because of the automation of the calculation and the user-friendliness of Expert Choice. With the sensitivity analysis, the participants were given the opportunity to check the reasonableness and robustness of the results.

a) Priorities calculation

The global priorities can be calculated in aggregating: the weighting schema for the stakeholder groups (tables 2 and 3), the importance of criteria (tables 4 and 5) and the local priorities of the alternatives with respect to the criteria (e.g. tables 6 and 7). The global priorities can be seen in table 8 and 9.

Strategic alternatives Priorities
Plant Redesign 0.373
Plant Automation 0.142
Plant Consolidation 0.485 Table 8: Global priorities of strategic alternatives resulted from cost analysis.

Strategic alternatives Priorities
Plant Redesign 0.277
Plant Automation 0.496
Plant Consolidation 0.227 14

Table 9: Global priorities of strategic alternatives resulted from benefit analysis.

Plant Plant Plant
Consolidation Redesign Automation
0.227 7 7 0.277 70.496 ~ Benefit
0
/ J I
/ J I
/ J I
/
/ J I
/ J J
Cost
0.485 0.373 0.142 0
Figure 8: Graphical representation of the benefits and costs analysis. From figure 8, it can be concluded that, the lowest cost option is 'Plant Automation' since it scored the least and the option with the most benefits is also 'Plant Automation' as it scored the highest in the benefit analysis. There is no incomparability in this problem as the cost and benefit analysis lead to the same ranking. Therefore, it is without surprise that the 'Plant Automation' is the most preferred alternative according to the results of the cost and benefit analysis (table 10).

Strategic alternatives Costs Benefits Benefit/Cost
Plant Redesign 0.373 0.277 0.74
Plant Automation 0.142 0.496 3.49
Plant Consolidation 0.485 0.227 0.47 Table 10: Cost-Benefit Analysis.

b) Sensitivity analysis

Figures 9 and 10 show the results of the performance of the three strategic alternatives based on the stakeholders' criteria. We can see the priority of each alternative and the weights of each Shareholder. In both figures, we can see that the 'Plant Automation' is the preferred alternative for each stakeholder. A 'What-if analysis will therefore not change the final result, which is a robust one.

Ob'%
.90
.80
.70
.60
.50
.40
.30
.20
.10 Alt%.60

Plant Automation

Plant Redesign

Plant Consolidation

15

00 .00

Shareholders Senior Mgrs Middle Mgrs OVERALL

Figure 9: Performance sensitivity of the strategic alternatives from the benefits analysis. The left vertical axis represents the weight of the stakeholders and the right vertical axis gives the priority of each alternative.

Alt% .60

.90

.80

Plant Consolidation

.70

Plant Redesign

.50

..to

.30

.20

Plant Automlltioll

.10

OOL....l....------'-'-----___JL....1....--------'----------'.OD

Share holder Senior Mgrs Middle Mgrs OVERALL

Figure 10: Performance sensitivity of the strategic alternatives from the costs analysis. The left vertical axis represents the weight of the stakeholders and the right vertical axis gives the priority of each alternative.

8. Results of the implementation

The recommendations of the model have been implemented with the general satisfaction of all stakeholders.

The successful acceptance of the proposed methodology can be attributed to the following reasons. Firstly, it helped to describe the problem and break down decision criteria into manageable components. Secondly, it led the group into making a specific decision for consensus or tradeoff. Thirdly, it provided an opportunity to examine disagreements and stimulate discussion and opinion. Fourthly, the process offered an opportunity to perform a sensitivity analysis in modifying judgments. Finally, it made possible to incorporate conflicts in perceptions and in judgments in the model.

The successful implementation of the recommendations of the model in this case study has empirically demonstrated the validity of the process and the GAHPO method.

16

9. Conclusions

In this paper we have presented the GAHPO a new multi-criteria decision aid method developed to solve a real problem. The backbone of the method is the AHP with several improvements:

Cost and benefit criteria are separated in two hierarchies in order to simplify their compansons.

Stakeholders are incorporated in the first level of the hierarchy in order to elicit a group preference.

The weight of each stakeholders are determined by others stakeholders. A consistency check is applied in order to verify the coherence of the comparisons given by the appraisers. A veto possibility is given to each evaluated stakeholder.

Two rankings exist: a partial ordinal ranking and a complete cardinal ranking. A partial ordinal ranking incorporates three possible relations: the preference) indifference and incomparability relation. The complete cardinal ranking fully aggregates the cost and benefit analysis.

The result of the proposed methodology showed that out of the three strategies, plant automation was the most preferred alternative. All participants were completely satisfied from this robust result. However before the adoption) a traditional financial analysis (discounted cash flow) has been conducted in order to assess the profitability of the selected alternative. In fact) the cost-benefit analysis with the GARPO ranks the alternatives but there is no guarantee that they will generate profits (Wedley et al., 2001). The first ranked alternative could be simply the one with the least loss.

By applying the methodology as a cost-benefit analysis) four main benefits have been achieved by the decision makers:

1. Significant reduction of time and effort in the decision process due to a structured methodology;

2. Easiness for the decision makers to arrive at a consensus) because the hierarchy model brings a common reference) which can be debated;

3. Enhancement of the decision quality) due to the consistency check and sensitivity analysis

embedded in the GARPO method;

3. Documentation and justification ofthe decision made.

The proposed GARPO methodology could be easily applied to other strategic selection problems, where several stakeholders are involved.

17

References

Aczel, J.) & Saaty, T. (1983). Procedures for synthesizing ratio judgements. Journal 0/ Mathematical Psychology, 27(1)) 93-102.

Altuzarra, A., Moreno-Jimenez, J., & Salvador, M. (2007). A Bayesian priorization procedure for AHP-group decision making. European Journal of Operational Research, 182(1),367- 382.

Amiri, M.) Zandieh, M., Soltani, R, & Vahdani, B. (2009). A hybrid multi-criteria decisionmaking model for firms competence evaluation. Expert Systems with Applications, 36(10), 12314-12322.

Arbel, A, & Orgler, Y. (1990). An application of the ARP to bank strategic planning: The mergers and acquisitions process. European Journal of Operational Research, 48(1),27-37.

Azis, 1. (1990). Analytic Hierarchy Process in the benefit-cost framework: A post-evaluation of the Trans-Sumatra highway project. European Journal of Operational Research, 48(1), 38-48.

Cebeci, U. (2009). Fuzzy ARP-based decision support system for selecting ERP systems in textile industry by using balanced scorecard. Expert Systems with Applications, 36(5),8900- 8909.

Celik, M., Kandakoglu, A., & Er, D. (2009). Structuring fuzzy integrated multi-stages evaluation model on academic personnel recruitment in MET institutions. Expert Systems with Applications, 36(3, Part 2),6918-6927.

Chamodrakas, I., Batis, D., & Martakos, D. Supplier selection in electronic marketplaces using satisficing and fuzzy AHP. Expert Systems with Applications, In Press, Corrected Proof, doi:1O.1016/j.eswa.2009.l 005.1 043.

Chan, S. T. F., & Chan, K. H (2004). Development of the supplier selection model- a case study in the advanced technology industry. Proceedings of the Institution of Mechanical Engineers. Part B. Journal of engineering manufacture, 218(12), 1807-1824.

Chang, C.-W., Wu, C.-R., & Lin, H-L. (2009). Applying fuzzy hierarchy multiple attributes to construct an expert decision making process. Expert Systems with Applications, 36(4), 7363-7368.

Chen, M. K., & Wang, S.-C. (2009). The critical factors of success for information service industry in developing international market: Using analytic hierarchy process (AHP) approach. Expert Systems with Applications, In Press, Corrected Proof,

doi: 10.1 016/j .eswa.2009.1 006.1 012.

Cheng, E., Li, H., & Ho, D. (2002). Analytic Hierarchy Process (AHP): a defective tool when used improperly. Measuring Business Excellence, 6(4), 33-37.

Cho, Y.-G., & Cho, K.- T. (2008). A loss function approach to group preference aggregation in the AHP. Computers & Operations Research, 35(3), 884-892.

Clayton, W., Wright, M., & Sarver, W. Benefit cost analysis of riverboat gambling. Mathematical and Computer Modelling, 17(4-5), 187-194.

Condon, E., Golden, B., & Wasil, E. (2003). Visualizing group decisions in the analytic hierarchy process. Computers & Operations Research, 30(10), 1435-1445.

Craven, J. (1992). Social Choice: A Framework/or Collective Decisions and Individual Judgements. Cambridge: Cambridge University Press.

Dagdeviren, M., Yavuz, S., & Killnc, N. (2009). Weapon selection using the AHP and TOPSIS methods under fuzzy environment. Expert Systems with Applications, 36(4),8143- 8151.

Forman, E., & Peniwati, K. (1998). Aggregating individual judgments and priorities with the analytic hierarchy process. European Journal 0/ Operational Research, 108(1), 165-169.

Ho, W., & Emrouznejad, A. (2009). Multi-criteria logistics distribution network design using SAS/OR. Expert Systems with Applications, 36(3, Part 2), 7288-7298.

Hsu, S. H., Kao, C.-H., & Wu, M.-C. (2009). Design facial appearance for roles in video games. Expert Systems with Applications, 36(3, Part 1),4929-4934.

Hsu, Y.-L., Lee, C.-H., & Kreng, V. B. The application of Fuzzy Delphi Method and Fuzzy AHP in lubricant regenerative technology selection. Expert Systems with Applications, 111 Press, Corrected Proof, doi:1 0.1 016/j.eswa.2009.1 005.1068.

Hurley, W. L, & Lior, D. U. (2002). Combining expert judgment: On the performance of trimmed mean vote aggregation procedures in the presence of strategic voting. European Journal 0/ Operational Research, 140(1), 142-147.

19, Y. T., & Yurdakul, M. (2009). Development of a decision support system for machining center selection. Expert Systems with Applications, 36(2, Part 2),3505-3513.

Jaganathan, S., Erinjeri, J., & Ker, J. (2007). Fuzzy analytic hierarchy process based group decision support system to select and evaluate new manufacturing teclmologies. The International Journal of Advanced Manufacturing Technology, 32(11), 1253-1262.

18

Khosla, R, Goonesekera, T., & Chu, M.-T. (2009). Separating the wheat from the chaff: An intelligent sales recruitment and benchmarking system. Expert Systems with Applications, 36(2, Part 2),3017-3027.

Labib, A, & Shah, J. (2001). Management decisions for a continuous improvement process in industry using the Analytical Hierarchy Process. Journal of Work Study, 50(5), 189-193.

Labib, A, Williams, G., & O'Connor, R (1996). Formulation of an appropriate productive maintenance strategy using multiple criteria decision making. Maintenance Journal, 11(11), 66-75.

Lai, W.-H., & Tsai, C.- T. (2009). Fuzzy rule-based analysis of firm's technology transfer in Taiwan's machinery industry. Expert Systems with Applications, 36(10), 12012-12022.

Li, S., & Li, J. Z. (2009). Hybridising human judgment, AHP, simulation and a fuzzy expert system for strategy formulation under uncertainty. Expert Systems with Applications, 36(3, Part 1),5557-5564.

Li, T.-S., & Huang, H.-H. (2009). Applying TRIZ and Fuzzy AHP to develop innovative design for automated manufacturing systems. Expert Systems with Applications, 36(4),8302- 8312.

Limam Mansar, S., Reijers, H., & Ounnar, F. (2009). Development of a decision-making strategy to improve the efficiency ofBPR. Expert Systems with Applications, 36(2, Part 2), 3248-3262.

Liu, C.-C., & Chen, S.- Y. (2009). Prioritization of digital capital measures in recruiting website for the national armed forces. Expert Systems with Applications, 36(5),9415-9421.

Madu, C., & Kuei, C.-H. (1995). Stability analyses of group decision making. Computers & Industrial Engineering, 28(4), 881-892.

Mikhailov, L. (2004). Group prioritization in the AHP by fuzzy preference programming method. Computers & Operations Research, 31(2),293-301.

Naghadehi, M. Z., Mikaeil, R, & Ataei, M. (2009). The application of fuzzy analytic hierarchy process (F AHP) approach to selection of optimum underground mining method for Jajarrn Bauxite Mine, Iran. Expert Systems with Applications, 36(4),8218-8226.

Niaraki, A. S., & Kim, K. (2009). Ontology based personalized route planning system using a multi-criteria decision making approach. Expert Systems with Applications, 36(2, Part 1), 2250-2259.

O'Leary, D. (1993). Determining Differences in Expert Judgment: Implications for Knowledge Acquisition and Validation. Decision Sciences, 24(2),395-408.

Onut, S., Efendigil, T., & Saner Kara, S. (2009). A combined fuzzy MCDM approach for selecting shopping center site: An example from Istanbul, Turkey. Expert Systems with Applications, In Press, Uncorrected Proof, doi:l0.l016/j.eswa.2009.1006.1080.

Pan, N. (2009). Selecting an appropriate excavation construction method based on qualitative assessments. Expert Systems with Applications, 36(3, Part 1),5481-5490.

Ramanathan, R, & Ganesh, L. (1994). Group preference aggregation methods employed in AHP: An evaluation and an intrinsic process for deriving members' weightages. European Journal of Operational Research, 79(2),249-265.

Saaty, T. (1977). A scaling method for priorities in hierarchical structures. Journal of mathematical psychology, 15(3), 234-281.

Saaty, T. (1980). The Analytic Hierarchy Process. New York: McGraw-HilL

Saaty, T., & Shang, J. (2007). Group decision-making: Head-count versus intensity of preference. Socio-Economic Planning Sciences, 41(1),22-37.

19

Saaty, T., & Vargas, L. (2007). Dispersion of group judgments. Mathematical and Computer Modelling, 46(7-8),918-925.

Saaty, T. L., & Vargas, L. G. (2005). The possibility of group welfare functions. International Journal of Information Technology & Decision Making, 4(2), 167-176.

Secme, N. Y., Bayrakdaroglu, A, & Kahraman, C. (2009). Fuzzy performance evaluation in Turkish Banking Sector using Analytic Hierarchy Process and TOPSIS. Expert Systems with Applications, 36(9), 11699-11709.

Sen, C. G., & Clnar, G. (2009). Evaluation and pre-allocation of operators with multiple skills: A combined fuzzy AHP and max-min approach. Expert Systems with Applications, In Press, Corrected Proof, doi: 1 0.1 0 16/j .eswa.2009.1 006.1 075.

Srdjevic, B. (2007). Linking analytic hierarchy process and social choice methods to support group decision-making in water management. Decision Support Systems, 42(4),2261-2273.

Tseng, Y.-F., & Lee, T.-Z. (2009). Comparing appropriate decision support of human resource practices on organizational performance with DEAl AHP model. Expert Systems with Applications, 36(3, Part 2),6548-6558.

Van Den Honert, R., & Lootsma, F. (1997). Group preference aggregation in the multiplicative AHP The model of the group decision process and Pareto optimality. European Journal a/Operational Research, 96(2),363-370.

Vidal, L.-A., Sahin, E., Martelli, N., Berhoune, M., & Bonan, B. (2009). Applying AHP to select drugs to be produced by anticipation in a chemotherapy compounding unit. Expert Systems with Applications, In Press, Corrected Proof, doi: 10.l016/j.eswa.2009.l 006.1067.

Wang, H. S., Che, Z. H., & Wu, C. Using analytic hierarchy process and particle swarm optimization algorithm for evaluating product plans. Expert Systems with Applications, In Press, Corrected Proof, doi: 10.10 16/j .eswa.2009.1 005.1073.

Wang, T.-Y., & Yang, Y.-H. (2009). A fuzzy model for supplier selection in quantity discount enviromnents. Expert Systems with Applications, 36(10), 12179-12187.

Wedley, W., Choo, E., & Schoner, B. (2001). Magnitude adjustment for AHP benefit/cost ratios. European Journal of Operational Research, 133(2),342-351.

Weiss, E., & Rao, V. (1987). AHP design issues for large-scale systems. Decision Sciences, 18(1),43-57.

Wu, C.-R., Lin, C.-T., & Lin, Y.-F. (2009). Selecting the preferable bancassurance alliance strategic by using expert group decision technique. Expert Systems with Applications, 36(2, Part 2),3623-3629.

Xie, G., Zhang, J., Lai, K., & Yu, L. (2008). Variable precision rough set for group decisionmaking: An application. International Journal 0/ Approximate Reasoning, 49(2),331-343.

Yang, C.-L., Chuang, S.-P., & Huang, R-H. (2009). Manufacturing evaluation system based on AHP/ANP approach for wafer fabricating industry. Expert Systems with Applications, 36(8), 11369-11377.

20