You are on page 1of 15

This article was downloaded by: [UNICAMP]

On: 18 June 2015, At: 07:08


Publisher: Routledge
Informa Ltd Registered in England and Wales Registered Number:
1072954 Registered office: Mortimer House, 37-41 Mortimer Street,
London W1T 3JH, UK

Journal of Environmental
Planning and Management
Publication details, including instructions
for authors and subscription information:
http://www.tandfonline.com/loi/cjep20

Applying Multi-attribute
Utility Theory to Decision
Making in Environmental
Planning: A Case Study
of the Electric Utility in
Korea
Tai-Yoo Kim , Seung-Jun Kwak & Seung-Hoon
Yoo
Published online: 02 Aug 2010.

To cite this article: Tai-Yoo Kim , Seung-Jun Kwak & Seung-Hoon Yoo
(1998) Applying Multi-attribute Utility Theory to Decision Making in
Environmental Planning: A Case Study of the Electric Utility in Korea,
Journal of Environmental Planning and Management, 41:5, 597-609, DOI:
10.1080/09640569811470
To link to this article: http://dx.doi.org/10.1080/09640569811470

PLEASE SCROLL DOWN FOR ARTICLE


Taylor & Francis makes every effort to ensure the accuracy of
all the information (the Content) contained in the publications
on our platform. However, Taylor & Francis, our agents, and our
licensors make no representations or warranties whatsoever as to the
accuracy, completeness, or suitability for any purpose of the Content.
Any opinions and views expressed in this publication are the opinions
and views of the authors, and are not the views of or endorsed by
Taylor & Francis. The accuracy of the Content should not be relied
upon and should be independently verified with primary sources of
information. Taylor and Francis shall not be liable for any losses,

actions, claims, proceedings, demands, costs, expenses, damages,


and other liabilities whatsoever or howsoever caused arising directly
or indirectly in connection with, in relation to or arising out of the use
of the Content.

Downloaded by [UNICAMP] at 07:08 18 June 2015

This article may be used for research, teaching, and private study
purposes. Any substantial or systematic reproduction, redistribution,
reselling, loan, sub-licensing, systematic supply, or distribution in any
form to anyone is expressly forbidden. Terms & Conditions of access
and use can be found at http://www.tandfonline.com/page/termsand-conditions

Journal of Environm ental Planning and M anagem ent, 41(5), 597 609, 1998

A pplying M ulti-a ttribute U tility T heory to D ecision


M a king in Environm ental Pla nning: A C a se Stu dy of
the Electric U tility in K orea

T A I-YO O KIM *, SE U N G-JUN KW AK & SE U N G-H OO N YO O*


*T echno-econo m ics a nd Po licy Pro gram m e, Seo ul N ationa l U niv ersity , San 56 1 ,
Sh inrim -D ong , K w a nak -Ku, Seo ul, 1 51 7 4 2 , K o rea .

Downloaded by [UNICAMP] at 07:08 18 June 2015

D ep a rtm ent o f Econom ics, Ko rea U niv ers ity , 5 1 A nam -D ong , Sung buk -K u, Seoul, 1 3 6 70 1 ,
K o rea
(R eceiv ed February 1 9 9 8; rev ised June 1 9 9 8 )

A BSTRACT This paper applies the m ulti-attribute utility th eory (M A U T) to obtaining


value judgm ents concerning the signi cance of environm enta l im pacts to achieve
integration of env ironm ental concerns at an early stage of planning , using the electric
utility of K orea as a speci c case stud y. A n environm enta l m ulti-attribute index is
constructed as a m ulti-attribute utility function, based on value judgm ents provided by
a group of experts related to electric utility and a d ecision m aker from K orea Electric
Pow er Corp oration (KEPC O ). The societal values are derived from exam ining trade-offs
betw een env ironm enta l index and m oney. The im plications of the results for K EPCO are
also discussed. W e foun d that the w ork and results can provide valuable insights and
decision opportunities for m ajor decision m aking in env ironm ental plann ing facing
K EPCO .
Introduction
Sin ce the 1992 Earth Sum mit in Rio de Janeiro, the env iron mental and social
impacts of en erg y developm en t have becom e an increasin gly im portant topic of
public deb ate in developing countries, such as Korea, as w ell as in developed
countries. In this general climate of increasing scrutiny of the en vironm en tal
implications of the development process, the electric utilities have received
particular attention, as a conseq uence of study results that associated the utilities
w ith signi cant potential env iron mental im pacts (H ohm eyer & Ottinger, 1994).
Even if it is uncertain w hether suf cient scien ti c evid en ce for the impacts
exists, there is general agreem en t that the most pressin g need is incorporating
en viron mental concerns proactively into investm en t plann ing decision making
rather than sim ply reacting to en viron men tal problem s after they occur. In
particular, the en vironm en tal options faced by Korea Electric Power C orporation
(KEPC O) becom e m ore complex and need to be more appropriately address ed
since KEPC O is resp onsible for 14% of total national air pollutant emissions and
is sued for en viron mental problem s such as marine products loss by therm al
discharge w ater. This situation clearly asks resea rchers to provid e decision
0964-0568/98/050597-13 $7.00

1998 Un iversity of N ewcastle upon T yn e

Downloaded by [UNICAMP] at 07:08 18 June 2015

598

T.-Y . K im , S.-J. Kw ak & S.-H . Y oo

makers (DM s) w ith available and respon sible information regard ing decision
making in en vironm en tal plann ing .
Recen tly, m any studies to value en viron men tal im pacts have been applied to
social costin g for these kind s of decision making (for example, Ottinger et al.,
1990; Thayer et al., 1994; D esvousges et al., 1995; European Commission , 1995;
Row e et al., 1995). These techniques have been successful in seeking to demonstrate how important incorporating env iron mental concern s into the plann ing
stage of electric utility developm en t is and in perm ittin g some impacts to be
folded into a conven tional cost ben e t analysis (H anley & Spash, 1994). H ow ever, som e researchers suggest that the valuation techn iques have room for
improvem en t (Gregory et al., 1993; Baron, 1996) and, for many of the most
important electric utility im pacts, they are very dif cult to apply.
This is due to tw o reasons . The rst is the nature of the im pacts themselves.
The en vironm en tal im pacts of electric utility such as health effects, agricultural
crop yield loss, visibility impairm ent, erosion and soil dam age, and global
w arm ing are extrem ely dif cult to monetize in term s of time and m oney.
Indeed , attempts to do so w ould be very lik ely to focus attention on the valid ity
of the valuation techn iques themselves , rather than the policy trade-offs that
must be m ade (Meier & M unasin ghe, 1993).
The second reason concern s the scale of analysis. The techniques for economic
valuation of en vironm en tal extern alities is usually m ost appropriate to the
project level. For example, the use of contingent valuation method (N ational
O ceanic and A tmospheric Adm inistration (N O AA ), 1993), w hich is the most
w idely used , is much more valid w here resp onden ts can be question ed on the
speci c im pacts of a speci c project to w hich they can relate. H ow ever, this may
be very lim ited and dif cult to apply in the context of long rang e utility
planning, w hich is dealin g w ith a potentially large number of technology , site
and m itigation options. Thus, the need s for en vironm en tal impact measures are
increasing ly expressed in term s of som e type of a unit value or value index
(Lin dsey et al., 1997). These need s require a change in the focus of research aw ay
from the speci cally de ned res ources to framew orks that seek to iden tify the
attrib utes of resources w hich distin guish their contrib utions to unit value
function.
It is in such situations that this study employs the M AUT (Multi-attrib ute
U tility Theory) elicitation technique as a basis for obtaining value judgm en ts
reg ard in g the sign i cance of env iron ment im pacts by usin g the electric utility of
Korea as a case study. The paper proceeds as follow s. Firs t, an overview of the
proposed method ology is presente d; second, the case study is explained, draw ing on interview s and the M AU T elicitation w ith a D M from KEPC O; third , the
implications of the results and decision opportunities for KEPCO are discussed ;
conclusion s appear in the nal section.
M ethodo logy: M A U T
The M A UT approach is rm ly groun ded in theory, speci cally the utility theory
of von N eumann & M orgenstern (1947), and the speci c assessm en t techn iques
and elicitation procedures are develop ed by Keen ey & Raiffa (1976). The
techn iques allow for in dividual-speci c attitudes about risk and selection of
attrib utes to be incorporated into the m odel. In addition, iden ti cation of the
appropriate functional form of the model is axiomatically derived from testable

Downloaded by [UNICAMP] at 07:08 18 June 2015

A pplying M ulti-attribute U tility Theory to D ecision M aking

599

beh avioural assum ptions concerning preferen ces and utility ind epend ence. The
attitudes, risk attitudes, trade-offs among attributes and com bination rules are
all determ ined through special procedures and questions asked of the D M .
Its application to public utility plannin g problem s has evolved steadily over
the last 20 years. A s the intangible en viron mental externalities lying outside
convention al cost ben e t analysis method ologies becom e increasin gly recognized, they have often been applied in electric utilities contexts. For exam ple,
Keen ey et al. (1985) analyse altern atives for regu lating air emissions from
coal- red pow er plants; M eier & M unasingh e (1993) assess the env iron mental
impacts of several gen erating sources for decision makin g concern ing Sri
Lank as electric pow er sector; the U S Energ y Information Administra tion (EIA )
(1995) consid ers it as one of the ve alterna tives to treat en viron men tal extern alities.
The paper that m ay be closest in spirit to our ow n is the recen t paper by
M cD aniels (1996). H e structures and quanti es basic values of env iron mental
impacts of the Britis h Columbia H ydro in C anada, by focusin g on a part of
Keen ey & M cDaniels (1992) that addres ses a w ide array of electric sys tem
strategic plannin g question s. H e does not provid e suf ciently the com plete
form s of single attribute utility functions nor for the practical uses of the analysis
to the utility, w hich are the foci of our paper. Our paper attempts to provide a
more careful consid eration of the en viron men tal im pacts of electric utilities to
explore the practicality of usin g M AU T in a developing country settin g w hen the
economic valuation of en viron men tal im pacts is very dif cult. The message of
our paper is all the more useful because there have been few develop ing w orld
applications of M AU T.

A pplication to K EPC O
O ur main concern is to construct an ind ex that w ould speci cally represen t the
env iron mental effects as an overall utility function or multi-attrib ute utility
function (MUF), nam ely, a mathem atical represen tation of value priorities and
trade-offs. The DM of our study w as the Vice-D irector of Electric Resources
1
Planning of KEPC O. W hile num erous M AUT procedures exist in theory and
practice (von W interfeldt & Edw ard s, 1986), this study adopts seven steps.

Step 1: Identifying O bjectives and A ttributes


A prelim inary set of objectives and attrib utes w as deriv ed through an extensive
foreig n and dom estic literature rev iew on en viron men tal impacts of electric
utilities. This set w as then thorough ly review ed and revised through a feedback
process by a group of 30 experts made up of policy analysts, resea rchers,
faculties and the DM . Durin g the process, a series of interview s w ere conducted
w ith practitioners from severa l divis ions of KEPCO to presen t objectives and
attrib utes that they though t should be crucial for the D M. N aturally, these
interview s en abled various objectives and attributes w hich include all aspects of
env iron mental im pacts of KEPCO to be identi ed. H ow ever, instead of attempting to equally and fully study all of the impacts, w e and the D M use the
screen ing process to decid e a nal set of objectives and attrib utes. The screening
criteria are as follow s:

Downloaded by [UNICAMP] at 07:08 18 June 2015

600

T.-Y . K im , S.-J. Kw ak & S.-H . Y oo

(1) Environ mental impacts sh ould be extern alities. Some im pacts associated
w ith electric resource options may not be extern alities. For example, dam age
to pow er plant w ork ers caused by accidents may be partially or w holly
intern alized in electricity production costs through insurance or w age premium. Such potential damages are deleted from the study.
(2) The externality sh ould be variable. Som e im pacts may be extern alities, but
may be xed for electric resource decisions. For example, the aesthetic
impacts of transmission and distribution facilities do not chang e w ith
selection of different electric resource options. Such extern alities are listed
unim portant and no study effort is recomm en ded to be undertaken .
(3) The literature should be suf cient. For some im pacts, the scienti c literature
may indicate that damage may happen but there is insuf cient in formation
to compute w ith any con dence the impact or damage at the location of
interes t. For exam ple, it may be appropriate to leave health effects from
nuclear pow er plant and biodiversity loss for future study since these
impacts are uncertain.
(4) Damage sh ould not be relatively small. Some ex ternalities, w hich m ay be
judged to be small relative to other externalities consid ered in the study
because of emission control and resource use reg ulations, are elim in ated in
the nal set of attributes. For example, coal ash can contaminate ground
w ater and cause aesthetic im pacts from trucking and the pres en ce of land ll
sites. H ow ever, KEPCO is recycling a substantial quantity of ash (21.9% in
1996) for concrete and other products, thereb y red ucing the amounts disposed of in land lls and other facilities, and paying consid erable money for
private disposal services for non-recycled ash .
(5) The extern alities sh ould take place frequently. Som e extern alities may be
associated w ith electric res ource options that constitute a sm all portion of the
capacity m ix in KEPC O. For example, w ind and solar-pow ered electric
resources are each forecast to rem ain at les s than 1% of KEPCO for the next
tw o decades. Therefore, study ing the im pact on such resource options may
in uen ce only a few decisions that affect only a small portion of electricity
supplies.
(6) The im pacts sh ould be of direct importance to national D Ms. Some impacts
may be externalities in the perspective of fuel life cycle, but they may be less
important relative to other impacts in national plann ing. In these cases,
additional study efforts may be unnecessary. For example, the en viron men tal im pacts associated w ith the extraction of imported coal that w ould arise
in A ustralia should be included in the en viron men tal costs for Australian
policy m aking , w hich w ould sim ply be re ected in higher coal prices to
KEPCO .

The overall objective for KEPC O plannin g is to comply w ith the publics
env iron mental values by electric resource planning to m inimize adverse environ men tal impacts of electric utility . This objective is, in turn , classi ed into
three sub-objectives: local en vironm en tal impacts; health effects; and global
w arm ing . Local en vironm en tal impacts are further classi ed into six attributes
and health effects into tw o as lis ted in Table 1. C onseq uently, the number of
attrib utes, w hich should be given particular attention, am ounts to nine. This is
quite appropriate, considerin g that trade-offs w ould become dif cult to under-

A pplying M ulti-attribute U tility Theory to D ecision M aking

601

T able 1. Objectives and attributes


1. T o
1.1
1.2
1.3
1.4
1.5
1.6
2. T o
2.1
2.2
3. T o

m inimize local en vironmen tal im pacts


T o for est
T o agricultural produ ction
T o m aterial
T o visibility
A b out therm al disch arge w ater
T o land use
m inimize detrim ental h ealth effects
T o public m ortality
T o public m orbidity
m inimize global w arm ing im pacts

Downloaded by [UNICAMP] at 07:08 18 June 2015

stand and display to the D M in a compreh en sible form if there are too many
attrib utes (Phelp s & Shanteau, 1978).
Step 2: Q uantifying A ttributes
In de ning the levels of attributes, a w ide range of background m aterials
produced by various sources w as collected. Es pecially as far as health effects are
concerned , w e concentrate on only domestic literature because adopting the
env iron mental and ambient stand ard s of the developed countries may imply a
seriou s misallocation of resources, consid ering that health effects play a consid erable part in en viron men tal im pacts (Row e et al., 1995). For each attrib ute,
w e specify the potential ranges of perform ance on the selective attributes, from
their best to their w orst levels, over the rang e of policies under consid eration.
The chosen ranges of attributes are presen ted in Table 2.
Step 3: V erifying the R elevance of A ssum ptions
This step derives the mathematical implications in M U F form . Th is sh ould be
based on tests of value ind epend ence composed of preferential indepen dence,
w eak-differen ce ind epen den ce, utility in depen den ce, and additive ind ependence. W hen four assumptions hold, an additive form is appropriate (Fish burn ,
1965). H ow ever, the question s of checking up indepen den ce are too complex and
time-consuming to w ork easily w ith a high-rankin g of cial. In practical term s,
an additive form sh ould be at least a good approximation to overall value.
A gain, w hen interview ing the sim plifying assumption of an additive structure is
almost alw ays employed to red uce the complexity of the elicitation procedures
(Edw ard s & von W interfeldt, 1987). The structure is also rob ust as a basis for
applied decision aiding in a variety of contexts and is the basic part of a more
com plicated multiplicative function. Furtherm ore, w e help ed the DM to provid e
ind epen den t value judgm en ts.
W e therefore assume that M UF has additive form as:
U (x 1 ,x 2 ,x 3 ,) 5

k 1 u 1 (x 1 ) 1

k 2 u 2 (x 2 ) 1

k 3 u 3 (x 3 )

(1)

w here U is the M UF containing three m ajor componen ts concernin g local


env iron mental impacts (x 1 ), health effects (x 2 ), and global w arm ing (x 3 ); k i s are
the scalin g constants on the major com ponen ts w hich satisfy

602

T.-Y . K im , S.-J. Kw ak & S.-H . Y oo


Table 2. Ranges and units of attrib utes
A ttributes
1.1

Forest:
area of for est lost w hich is ex pressed in
a
times of area of Yeoeu i-d o
A gricultural prod uction:
ann ual percen tage of agricultural
production loss
M aterial:
ann ual percen tage of m aterials dam ages
V isibility:
average d aily km of m ax im um
visibility in the m etropolitan area
T herm al d isch arge w ater:
ann ual percen tage of m arine produ cts loss
around thermal or n uclear pow er plants
L and use:
area of lan d required per M W of
electricity prod uced
Pub lic m ortality:
increase in ann ual n ew person s per m illion
of life lost due to lung cancer cases
Pub lic m orbidity:
increase in ann ual n ew person s per m illion
sufferin g from respiratory cases
Global w arm ing:
am oun t of electricity generated using fos sil fuels

1.2

1.3
1.4

Downloaded by [UNICAMP] at 07:08 18 June 2015

1.5

1.6

2.1

2.2

N otes :

i5

B est

Lev els
W orst

U n it

200

20

20

40

km

40

50

500

pyung

100

person

10 000

person

200 000

GW

Yeoeu i-do is a represen tative an d w ell-kn ow n island in Seoul, Korea; its area is ab out
b
2
300 h a. Pyu n g is a surface unit used in K orea; 1 pyun g 5 3.31 m .

ki 5

1,u i(x i),

is the objective utility function, and U and all the u i s are scaled from 0 to 1 over
the ranges in Table 2.

Step 4: Exam ining the Single A ttribute U tility Fu nction


In order to obtain com pletely the sing le attribute utility functions, rst, the
gen eral shape of them is determ ined by the risk attitudes such as risk aversion ,
risk neutrality and risk pronen ess. Second, the speci c utility function of that
gen eral sh ape is iden ti ed. The exponen tial and lin ear functional forms, w hich
are fairly robust sets of sin gle forms for characterizing single attrib ute utility
functions (Keeney, 1992), are assumed.
For all attributes, the D M sh ow ed consisten cies in risk assessm en ts need ed to
exam ine risk attitude and successfully accomplish ed certainty equivalent assessmen ts required to calculate risk parameters. Th e results in Table 3 show that the
attitudes tow ard forest, therm al discharge w ater, public m ortality and public
morb idity are risk-p rone, but those tow ard agricultural production, material,
visibility , land use and global w arm ing are risk- averse.

A pplying M ulti-attribute U tility Theory to D ecision M aking

603

T able 3. Res ults of sing le utility equations


A ttribute
x 11 5
x 12 5
x 13 5
x 14 5
x 15 5
x 16 5
x 21 5
x 22 5
x 35

U nit of c

Single attribute utility functions

for est
agricultural prod uction
m aterial
visibility
thermal d ischarge w ater
land use
public m ortality
public m orbidity
global w arm ing

u 11 (x 11 ) 5 2 2.26522 1 3.26522 exp( 2 1.84674 3 10 x 1 1 )


2 2
u 12 (x 12 ) 5 1.78406 2 0.78406 ex p(4.11073 3 10 x 12 )
2 2
u 13 (x 13 ) 5 1.38583 2 0.38583 ex p(6.39321 3 10 x 13 )
2 2
u 14 (x 14 ) 5 2.27407 2 2.47029 ex p( 2 1.65529 3 10 x 14 )
2 2
u 15 (x 15 ) 5 2 0.54301 1 1.54310 exp( 2 2.61065 3 10 x 1 5
2 3
u 16 (x 16 ) 5 1.30902 2 0.26322 ex p(3.20799 3 10 x 16 )
2 3
u 21 (x 21 ) 5 2 0.78406 1 1.78406 exp( 2 8.22163 3 10 x 2 1 )
2 5
u 22 (x 22 ) 5 2 2.26522 1 3.26522 exp( 2 3.69349 3 10 x 2 2 )
2 5
u 3 (x 3 ) 5 1.09574 2 0.09574 ex p(1.21876 3 10 x 3 )
3

%
%
km
%
pyung
person
person
GW

N o te s: Yeoeui-d o is a represen tative and w ell-kn ow n islan d in Seoul, K orea; its area is ab out 300 h a.

Downloaded by [UNICAMP] at 07:08 18 June 2015

Step 5: D eterm ining the Im portance of A ttributes


This step determ ines the DM s attitude tow ard the relative importance of the
various attributes by having him rank the attributes in order of im portance and
then give w eigh ts to them . The approach adopts a sw ing w eigh ting method,
w hich has been w idely employed and is the most defen sible w eig htin g scheme
(D ale et al., 1996). In order to check on the consis tency in resp onses, w e use tw o
versions of sw in g w eigh ting . The rst is to use the approach in the order of three
objectives, six attributes of local en vironm en tal impacts and tw o attributes of
health effects. The second is to apply it to all the nine attributes at once. For
ind epen den ce conditions , durin g the interview , the DM w as asked to bear in
min d that only the level of the attribute in question changes from the w orst to
the best but the others do rem ain at some origin ally speci ed level. These
procedures w ere w ell w ithin the D M s ability.
Step 6: C onstructing the M U F
The M UF can be constructed by usin g the information collected in steps 1 to 5
and nd ing single attribute functions and the scalin g constants k i s in equation
(1). The local en viron mental impacts utility function u 1 is assumed to be additive
w ithin six attrib utes and the health effects utility function u 2 is also additive
betw een public mortality and m orbidity. Thus, the nal MU F is:
U (x1 ,x 2 ,x 3 ) 5

5
1

U (x11 ,x 12 ,x 13 ,x 14 ,x 15 ,x 16 ,x 21 ,x 22 ,x 3 )
c 1 u 11 (x 11 ) 1
c 2 u 12 (x 12 ) 1
c 3 u 13 (x 13 ) 1
c 5 u 15 (x 15 ) 1
c 6 u 16 (x 16 ) 1
c 7 u 21 (x 21 ) 1

c 4 u 14 (x 14 )
c 8 u 22 (x 22 ) 1

c 9 u 3 (x 3 )

(1 9 )

w here U and all the u i s are scaled from 0 to 1 over the ranges in Table 2 and
the c i s are scalin g constants, w hich satisfy

i5

ci 5

and mean the relative importance of m oving the corresp onding attribute from its
w orst level to its best level.
The resultin g w eigh ts and ranking s are show n in the second and third
columns of Table 4. The coef cients clearly sugg est that the major strategic

604

T.-Y . K im , S.-J. Kw ak & S.-H . Y oo


T able 4. Assess ed priorities and value trade-offs of the attributes

A ttributes

Scaling con stants

Rank

0.0929
0.1032
0.0877
0.0722
0.0413
0.0619

4
3
5
6
8
7

32.5
365.0
310.0
145.7
72.5
9.8

0.2835
0.2267
0.0306

1
2
9

200.0
1.6
0.011

L ocal env ironm en tal im pacts (x 1 )


for est (x 1 1 )
agricultural prod uction (x 12 )
m aterial (x 13 )
visibility (x 14 )
therm al disch arge w ater (x 15 )
lan d use (x 16 )
H ealth effects (x 2 )
public m ortality (x 21 )
public m orbidity (x 2 2 )
Global w arm ing (x 3 )

Downloaded by [UNICAMP] at 07:08 18 June 2015

N ote:

Un it value
trade-offs in
a
w on

Ex pressed in terms of m on thly electric b ill per fam ily of fou r. At the time of the
interview , U S$1 is approx im ately equal to 800 K orean w on .

attrib utes that count concern public m ortality and morbidity, w ithin the ranges
de ned for the attrib utes. Global w arm ing explains only 3.06% of the en vironmen tal im pacts and is least important of the nin e attributes, given its range.

Step 7: A ssessing the Econo m ic V alue Trade-offs


W ith the rang es and units, w e had tw o w ays in think ing about the same issue
to check on the in consis tency. The 18 ( 5
2 3
9) questions are asked . Three
points are stresse d in the course of the in terview . First, w e are ask ing for
person al opinion s of an electric res ource developm en t policy DM rather than
organizational pers pectives to avoid the need for organizational approval and
agreem ent on judgments. Second, the measure w e try to value is fundam en tal or
societal value (Greg ory et al., 1993; Baron, 1996) instead of the w illin gness- to-pay
2
(W TP). Third , the values of interes t are those express ed by the DM and that
there are no righ t or w rong resp onses to the question s asked , although there are
interna lly applied consis tency criteria.
At rst, the D M s resp onses w ere not consis tent, but several feed back processes gave us consisten t res ponses. Th e fourth column of Table 4 presents the
res ults of the calculations to determ ine som e economic value trade-offs in term s
of the monthly electric bill per family of four. The results should be interpreted ,
for example, as follow s. One ann ual pers on-year of life lost is as bad as a 200
w on increase in the m onthly electric bill per fam ily of four, and 1 GW gen erated
using fossil fuels is as bad as 0.011 w on increase in the m onthly electric bill per
family of four.

Im plications of the R esults for K EPC O


Insights from the O bjectives and A ttributes
The objectives and attributes in Table 1 are useful for com munication and
information collection, particularly for electric resources plannin g and en viron-

Downloaded by [UNICAMP] at 07:08 18 June 2015

A pplying M ulti-attribute U tility Theory to D ecision M aking

605

men tal managem en t at KEPC O. O ne problem facing env iron mental plann ing in
KEPCO is insu f cient communications among the divisions. W e interv iew ed
severa l practitioners from the electric resources plannin g division , the en vironmen tal managemen t divisio n and the damage-com pensation division of KEPCO .
W e could ascertain that they do not possess a suf cient overview of the essen tial
strategic issues of other division s, w hich may be in uential in their decision s.
For instance, a practitioner from electric resources plannin g had no idea that the
damage-compensating division had been com pensating for agricultural production loss of the families doing farm w ork near the Youngd ong therm al pow er
plant w hich res very low -quality dom estic coal and em its a large am ount of
sulphur pollutants. H ow ever, the fact may affect decision makin g about new
pow er plant construction or sw itch of the plant fuel type from dom estic coal to
cleaner en erg y.
One of the most effective w ays to facilitate communication w ithin KEPC O is
to clarify objectives that are im portant for KEPC O s decision m aking in en vironmen tal plannin g and provid e all the division s w ith a structured set of objectives
and attributes, accompanied by an M UF derived above, w hich makes it easier
for them to make appropriate decisions w ithin their realms of resp onsib ility.
M oreover, such information w ould allow KEPC O to operate the en viron men tal options more appropriately . N ote from the scaling constants in Table 4 that
51.0% of the evaluations w ithin attribute ranges are due to health effects
objectives. Thus, this is likely to be a fruitful area in w hich to search for desira ble
alterna tives. It w ould probably be more appropriate to adopt an option to
red uce health effects than one to mitigate green house gas emission. Reg ulations
to address these health effects, so that KEPCO could ben e t from actions that
red uce such effects, m ay be useful.

Potential D ecision O pp ortunities


Each of the w orks consisting of value assessm ent w e presen t suggests potential
decision opportunities that may be w orth KEPC O pursuing. M any en viron men tally strategic decision s w ill face KEPC O in the next decade. These decision s
pertain to several issues: rst, pollution rem oval devices; second, the use of clean
fuels such as lique ed natural gas (LN G ); third , high ef cient and advanced
pow er generation technologies such as super-critical boilers, integrated
gasi cation combined cycle and uidized bed com bustion ; and fourth, R& D
inv estm en t on clean technology such as carbon dioxide recovery techn ology.
KEPCO w ill w ant to base such actions on good information and sound logic in
a co-ordinated m anner that can be justi ed and open ly com municated.
Other important decision opportunities involve technology choice issues.
These include: examining consum er decisions about demand -side managem ent
from both the economic and en viron men tal standpoints; clarifyin g low -quality
domestic stone coal (containing a larger amount of sulphur than im ported
bituminous coal) projects w hich are justi ed by the effect of the supply security
for gen erating fuels and the protection of domestic coal industry; and clarifying
strategies regard ing transmission and distribution loss (5.5% in 1995) red uction
w hich could abate emission s.

606

T.-Y . K im , S.-J. Kw ak & S.-H . Y oo

Downloaded by [UNICAMP] at 07:08 18 June 2015

C urrent U ses in K EPCO


KEPCO has used the analysis pres en ted in the previous sections in severa l
decision contexts during and since w e did this w ork, even though further efforts
could be need ed for many speci c decisions, in that it is not directly releva nt to
speci c projects or options . Speci cally, it is used in makin g long term pow er
developmen t plans to re ect the domestic and intern ational env iron mental
concerns and the reg ulations. This plan could make clear how new electric loads
w ill be served over the coming decade.
Firs t, KEPCO plans for a gradual increase in LN G capacity. On the other hand ,
the portion of oil- red and coal- red capacity w ill be red uced. Furtherm ore,
com bustion pow er plants are req uired to install ue gas desulphurization
(FGD), low -nitrate burn ers, electrostatic precipitators, selective catalytic reduction and other en viron men tal saving equipm en t. Increasing the use of low
sulphur fuel is also planned . Second, in order to promote the use of altern ative
energy res ources and clean coal techn ology (CC T), KEPCO plans to install 5 M W
of altern ative resource pow er plants by year 2003 and 300 M W of CC T pow er
plants by 2005. Third , by 2010, the carbon dioxide emission w ill be stabilized at
the level of 0.11 kg-C /kW h (0.2 kg-C/kW h in 1995) by extend in g the use of low
carbon emitted en erg y resources and improving the ef ciency.

Potential U ses of the U tility Fu nction


The utility function in Table 4 presents tw o potential uses in its application. First,
it provid es a value index, w hich could be used to calculate the env iron mental
losses at potential electric utility sites. If the DM has a set of alterna tive plans,
he may w ish to evaluate and rank that set in term s of positive env iron mental
impacts by m aking use of their associated total utilities. In other w ord s, he can
consid er altern atives of interes t for the multi-attribute utility and prefer one
w hose total utility is the largest. Second, asses sments of econom ic value can
provid e a basis to decide w hether to invest in pollution abatem en t actions. For
exam ple, from Table 4, a cost in crease of 20 000 w on on a m onthly electric bill
per fam ily of four m ay be view ed as equivalent to an improvem ent on the
env iron mental index of 0.2835. All opportunities to give rise to the improvem en t
of 0.2835 on the env iron mental impact ind ex at costs below 20 000 w on per
month over all custom ers w ould then be seen as a desira ble investm en t.

Fu rther D ecision O pp ortunities in A pplication


Furth er decision opportunities available to KEPCO need to be clari ed as a
second phase of the w ork in order to integra te en viron men tal consid erations into
the integrated resources plannin g of the electric utility of Korea. They pertain to
overcomin g obstacles to tw o potential uses discussed above. An important
obstacle in using the index in KEPC O is the general lack of detailed scien ti c
information available on speci c effects associated w ith electricity gen eration
and the resultin g emission s. For example, durin g the study , KEPC O w as
consid ering plans to inv est by 2005 in FG D facilities at most therm al pow er
plants. Their ann ual costs are estim ated at approximately 237.6 billion w on or
18.6 thousand w on annually per family of four. The plan focuses on air pollution
concentrations. One potential decision opportunity w ould be to examine

Downloaded by [UNICAMP] at 07:08 18 June 2015

A pplying M ulti-attribute U tility Theory to D ecision M aking

607

w hether it is socially desirable or not by usin g the index. H ow ever, w e have


dif culty in nding reliab le information on how the investm ent w ill improve the
ind ex, in a developing country settin g w here there is little prev ious res earch on
dose-response functions. Our study therefore suggests that a large number of
useful data, w hich could be used to kn ow the relationsh ips betw een air pollution
levels and the achievem en t of the attributes related to air pollution in Table 2,
sh ould be collected. It w ould also be useful to know w hat conseq uences in term s
of those attributes w ould be caused by measures to control air pollution.
The second obstacle is that w e focus on the view s of only one D M for
developing utility functions since w e are interes ted prim arily in the view points
of a KEPCO s DM as a national D M . From the KEPC O staff view point, the utility
must resp ond to the interests of a w ide variety of stakeh older groups because
KEPCO is a public utility and the only electric utility in Korea. The utility
functions re ecting the view s of various stakeh olders could improve understanding and env iron mental decisions by the utility. Thus, w e propose collecting
public opinion regard ing this issue in the near future. Under this plan, an appeal
w ill be made directly to citizenry for the evaluation of the policy option in
question . Such a fram ew ork may be desira ble sin ce any attempts to lim it
emissions may fail if there is not strong public support.
A dvanta ges of M A U T Technique
D uring and after the resea rch, w e and the DM found that the techn ique has
severa l strong advantages. First, it explicitly disting uish es the technical from the
value components of a decision, that is, experts or literature can provid e
techn ical inform ation, w herea s the D M or the stakeh olders can assess value
judgm en ts. Second, for value judgm ent, it asks societal values w hich are closely
connected w ith the problem , in stead of W TP that raises the question of the
distrib ution of incom e as a constrain t that effectively gives w ealthy people a
bigg er say in determ inin g a value (Gregory et al., 1993). Third , it does not need
holistic judgm ents, thus, it can accommodate the multi-dimens ionality of value
and allow the D M to evaluate several alterna tives w ith differen t env iron mental
impacts in term s of m inimization of adverse en viron men tal im pacts. Fourth, it
is very exible in changing circumstances, since it w ould elicit a broader range
of values for each attribute. Fifth , it coincides w ith one objective of m odern DM s,
w ho prefer to be presen ted w ith a range of feasib le alterna tives rather than one
best solution.
C onclusions
The express objective of this study w as to apply a methodology that can be
implem ented in practical w ays and w hose potential im pact on decision rules is
obviously dem onstrable. The approach used seeks to help analysts , practitioners,
and policy makers in the eld by developing practical procedures to assist
decision makin g and to presen t possible solutions to severa l of the most
troublin g problem s confronting en viron men tal resea rchers in developing countries. W e have also reported the im plications of the results for KEPC O.
Beyon d the intrinsic interests in relation to decision m aking in the en vironmen tal plann ing of KEPCO, this paper has demonstrated the feasibility of
extending the use of M AU T method s at least to a new ly industrialized A sian

608

T.-Y . K im , S.-J. Kw ak & S.-H . Y oo

nation. The sk ills developed in valuable previous M AU T studies applied to and


the in creasing public concern over en viron men tal impacts of electric utility
provid e a sound foundation on w hich to im pose the special requirem en ts of
M AU T study.

A cknow ledge m ents


W e acknow led ge the contributions of all in divid uals w ho participated in in terview s to identify objectives and attrib utes and the KEPCO s supports of our
w ork on this article and the concepts used in it. W e specially appreciate the
interes t and support of W oo-Chang C ho, the Vice-Director of Electric Resources
Planning of KEPC O. The authors are very grateful to Jeong -In Kim and an
anonymous referee for their helpful comm en ts.

Downloaded by [UNICAMP] at 07:08 18 June 2015

N otes
1.

2.

A lthough h e is a Vice-D irector, w e selected h im as a DM , since h e has w ider know led ge and
d eeper insigh ts into electric resources plann ing than the D irector, w h o h ad been appointed as
d irector for ve m on ths at the time of the interview .
Fund am en tal or societal value m eans h ow m uch society should pay to get a stated im provem en t
in en vironm en tal quality, w h ile the W T P m eans h ow m uch m y househ old or I w ould be w illing
to pay to d o so. Th e W T P of an ind ividual is like a m arket purch ase, w hile eliciting a view on
the overall value to society is like providing ad vice to policy m akers.

R eference s
B aron, I. (1996) Rationality and invarian ce: respon se to Sch um an, in: D . J. B jorn stad & J. R. K ahn (E d)
T h e C o ntin gent V a luatio n of En v iron m enta l Resou rces (C h eltenham , E d w ard E lgar).
D ale, V ., Russell, C ., H ad ley, M ., K an e, M . & Gregory, R. (1996) A pplying m ulti-attribute utility
tech niques to en vironm en tal valuation: a fores t ecos ystem stud y, prepared at the Southern
E con om ic A ssociation M eetings, N ovem b er (W ashington, D .C .).
D esvousges, W .H ., Joh nson , F.R., B an zhaf, H .S., Russell, R.R., Fries, E .E ., D ietz, K .J. & H elm s, S.C.
(1995) A ssessing E nv iron m enta l E xterna lity C osts fo r Electr icity G eneration , report to N orthern States
Pow er C om pan y (M inn esota) (D urham , N C , Triang le Econom ic Research).
E d w ards , W . & von W interfeld t, D . (1987) Pub lic values in risk deb ates, R isk A n a ly sis, 7, pp. 141 158.
E uropean C om m ission (1995) Ex te rna lities o f En erg y: Ex te rn E Pr o ject fo r th e D irecto rate G eneral X II,
M etroeco n om ia, C EP N , IE K , E y re E nerg y -E nv iron m ent (S tuttgart, E TS U , E cole d es M ines).
Fishb urn, P.C . (1965) In dep en d en ce in utility theory w ith w h ole prod ucts sets, O p eration R esearch, 13,
pp. 28 43.
Greg ory, R., L ichtenstein, S. & Slovic, P. (1993) V aluing en viron m ental resources: a con structive
approach, Jo u rnal of Risk & U n ce rtainty, 7, pp. 177 197.
H anley, N . & Spash, C .L . (1993) C ost B ene t A na ly sis a nd th e En v iron m ent (A ldershot, E d w ard Elgar).
H oh m ey er, O . & O ttinger, R.L . (E d ) (1994) S ocia l C osts of En erg y (Ber lin, Springer-Verlag).
K een ey , R.L . (1992) V a lue-F ocus ed T h inking (C am b ridge, M A , H arvard U n iversity Pr ess).
K een ey , R.L . & M cD aniels T .L . (1992) V alue-focu sed thinking about strategic d ecisions at B C H yd ro,
Inter fa ce s, 22, pp. 94 109.
K een ey , R.L . & Raiffa, H . (1976) D ecision s w ith M ultip le O bjectiv es: P reference s an d V alue T rad eoffs (N ew
York, John W iley).
K een ey , R.L ., Sicherm an , A . & Sm ith, G. (1985) A n alyzing radion uclide em issions fr om coal- red
pow er plants, In te rface, 15, pp. 12 24.
L ind sey, G., W ittman , J. & Rum m el, M . (1997) U sing ind ices in en vironmen tal plan ning: evaluating
policies for w ell eld protection, Jo urna l o f E n v iron m enta l Plan ning & M a na g em ent, 40, pp. 685 703.
M cD aniels, T .L . (1996) A m ultiattribute index for evaluating env ironm en tal im pacts of electric
utilities, Jo u rnal o f E n vironm en ta l M a n ag em ent, 46, pp. 57 66.
M eier, P. & M un ash ingh e, M . (1993) In corporating env iron m en tal con cern s into pow er sector
d ecision-m aking, W o rld B an k E n vironm ental P ap er N o. 6, (W ashington, D C , W orld Ban k).

A pplying M ulti-attribute U tility Theory to D ecision M aking

609

Downloaded by [UNICAMP] at 07:08 18 June 2015

N ational O cean ic and A tm osph eric A dm inistration (N O A A ) (1993) Report of the N O A A panel on
contingen t valuation, Federal R egister, 58, pp. 4602 4614.
O ttinger, R.L., W ooley , D .R., Robinson, N .A ., H oda s, D .R. & B abb , S.E . (1990) En v iron m enta l C o sts o f
E lectricity (N ew York, O cean a Pu b lications).
Ph elps, R.H . & Shan teau, J. (1978) L ivestock jud ges: how m uch inform ation can an ex pert use?,
O rga n iz ation a l B eh av io r & H um a n P erfo rm a nce, 21, pp. 209 219.
Row e, R.D ., L ang , C .M ., L atimer, D .A ., Rae, D .A ., B ern ow , S.M . & W h ite, D .E. (1995) N ew Y ork S ta te
E n vironm en ta l E x te rnalities C ost Stud y (N ew York, O ceana Pu blications).
T h ayer, M .A ., Seb old, F.D., M ayer, T .A ., M urdoch , J.C ., M urray, D .R., T racy, S.R. & Zapert, J. (1994)
T h e A ir Q ua lity V alua tion M o del, report to Ca lifornia E n ergy C om m ission (Sacram en to, C A )
Region al E con om ic Research (San D iego, C A ).
U S En ergy In form ation A d m inistration (E IA ) (1995) E lectricity gen eration and env iron m en tal extern alities: case studies, D O E/E IA -0 5 98 (W ash ington, D C , U S D epartm en t of E nergy).
von N eum ann , J. & M orgen stern, O . (1947) T h eory of G a m es a nd E co n om ic B eh av io r (Prin ceton , N J,
Pr inceton U n iversity Press).
von W interfeldt, D . & E dw ards, E. (1986) D ecisio n A n aly sis a nd B eh a viou ral R esearch (C am bridge,
C am b ridge Un iversity Press).

You might also like