Journal of Multi-Criteria Decision Analysis Volume 11 issue 4-5 2002 [doi 10.1002_mcda.328] Winfried...

7/25/2019 Journal of Multi-Criteria Decision Analysis Volume 11 issue 4-5 2002 [doi 10.1002_mcda.328] Winfried G. Hallerbac

1/9

JOURNAL OF MULTI-CRITERIA DECISION ANALYSIS

J. Multi-Crit. Decis. Anal. 11: 187195 (2002)

Published online in Wiley InterScience (www.interscience.wiley.com) DOI: 10.1002/mcda.328

The Relevance ofMCDM forFinancial Decisions

WINFRIEDG. HALLERBACH* andJAAPSPRONK

ErasmusUniversity Rotterdam, POBox1738,NL-3000DRRotterdam,TheNetherlands

ABSTRACT

For people working in finance, either in academia or in practice or in both, the combination of finance andmultiple criteria is not obvious. However, we believe that many of the tools developed in the field of MCDM cancontribute both to the quality of the financial economic decision making process and to the quality of the resultingdecisions. In this paper we answer the question why financial decision problems should be considered as multiplecriteria decision problems and should be treated accordingly. Copyright# 2003 John Wiley & Sons, Ltd.

KEY WORDS: MCDM; financial decision making; finance

1. INTRODUCTION

This special issue is devoted to MCDM andfinance. For people working in finance, either inacademia or in practice or in both, the combina-tion of finance and multiple criteria is notobvious. For many years, we have been involved inthe study and practice of financial economicdecision making. Our main focus was and is onfinancial management science, the development oftools and frameworks to support financial decisionmaking. We are convinced that many financialdecision problems are in fact multiple criteriadecision problems. In addition, we believe that

many of the tools developed in the field of MCDMcontribute both to the quality of the financialeconomic decision making process and to thequality of the resulting decisions. For examplesand overviews we refer to Martel and Zopounides(1998), Zopounides (1999), and Steuer and Na(2003). In this paper we answer the question whyfinancial decision problems should be consideredas multiple criteria decision problems and shouldbe treated accordingly.

The central issue in financial economics is theefficient allocation of scarce capital and resourcesover alternative uses. The allocation (and redis-

tribution) of capital takes place on financialmarkets and is termed efficient when marketvalue is maximized. Just as water will flow to the

lowest point, the capital will flow to the uses that

offer the highest return. Therefore it seems that thecriterion to guide financial decisions is uni-dimen-sional: maximize the market value or maximizefuture return.

From a financial economic perspective the goalof the corporate firm, for example, is very muchsingle objective. Management should maximize thefirms contribution to the financial wealth of itsshareholders. Also the shareholders are consideredto be myopic. Their only objective is to maximizetheir single dimensional financial wealth. The linkbetween the shareholders and the corporate firm isfooted in law. Shareholders are the owners of the

firm. They possess the property rights of the firmand are thus entitled to decide what the firmshould aim for, which in this line of thinking is thesame for all shareholders, i.e. maximization of thecontribution to the financial wealth of the share-holders. The firm can accomplish this by engagingin investment projects with positive net presentvalue. This is a neo-classical view on the role ofthe firm and the relationship between the firmand its shareholders in a capitalist society. Figure 1depicts a simplified graphical representation of thisline of thought.

It is important to note that this position is

embedded in a much larger framework of stylizedthinking in among others economics (generalequilibrium framework) and law (property rightstheory and limited liability of shareholders). Untiltoday, this view is seen as an ideal by many; seefor example Jensen (2001). Presently, however,the societal impact of the firm and its governancestructure is a growing topic of debate anddiscussion.

Received 15 May 2002Copyright# 2003 John Wiley & Sons, Ltd. Accepted 20 October 2002

*Correspondence to: Dr. W. Hallerbach, Departmentof Finance, Erasmus University Rotterdam, PO Box1738, NL-3000 DR Rotterdam, The Netherlands.E-mail: [email protected]


2/9

Here we will show that also in finance there aremany roads leading to Rome, or rather to thequasi Roman numeral MCDM. Whether onebelongs to the camp of Jensen or to the camp ofthose advocating socially responsible entrepre-neurship, one has to deal with multiple criteria.The following section is devoted to the position ofthe firm, while the third section addresses theposition of the financial investor. The Section 4discusses the issue of risk management that relatesto both the firm and the investor. Section 5concludes the paper.

2. MULTIPLE CRITERIA FINANCIALDECISIONS AND THE FIRM

In this section we describe a series of situations inwhich the firm chooses (or has to take account of)a multiplicity of objectives and (policy) con-straints. An overview of these situations isdepicted in Figure 2. One issue is who decides onthe objective(s) of the firm. If there is a multiplicityof parties who may decide what the firm is aimingfor, one generally encounters a multitude of goals,

constraints and considerations that}

more oftenthan not } will be at least partially conflictive. Aclear example is the conflicting objectives arisingfrom agency problems (Jensen and Meckling,1976). This means that the decision problems thathave to be solved are characterized by multiplecriteria and multiple actors (viz. group decisionmaking, negotiation theory, see Box 3 in Figure 2).Sometimes, all those who decide on what the firm

should aim for agree upon exactly the sameobjective(s). In fact, this is what neo-classicalfinancial theory assumes when adopting share-holder value maximization (Box 1 in Figure 2). Inpractice, there are many firms that explicitly strive

for a multiplicity of goals which naturally leads todecision problems with multiple criteria (Box 2 inFigure 2).

However, although these firms do explicitlystate to take account of multiple objectives, thereare still very few of these firms that make use oftools provided by the MCDM literature. In mostcases firms maximize one objective subject to(policy) constraints on the other objectives. Assuch there is nothing wrong with such a procedureas long as the location of these policy constraints ischosen correctly. In practice, however, one oftenobserves that there is no discussion at all about the

location of the policy constraints. Moreover, thereis often no idea about the trade-offs between thelocation of the various constraints and theobjective function which is maximized. In ouropinion, multiple criteria decision methodologiesmay help decision makers to gain better insights inthe trade-offs they are confronted with.

Now let us get back to the case in which theowner(s)/shareholders do have only one objectivein mind: wealth maximization. Although this is bydefinition the most prominent candidate for singlecriteria decision making, we will argue that even inthis case there are many circumstances in which

the formulation as a multiple criteria decisionproblem is opportune.

In order to contribute maximally to the wealthof its shareholders, an individual firm shouldmaximize the value of its shares. The value ofthese shares is determined on the financial marketsby the dynamic forces of demand and supply.Shares represent claims on the future residual cashflows of the firm (and also a usually very limitedright on corporate control). In the view of thefinancial markets, the value of such a claim isdetermined relative to the claims of other firmsthat are traded on these markets. The financial

markets perception of the quality of these cashflow claims is crucial for the valuation of theshares. Translated to the management of theindividual firm, the aim is not only to maximizethe quality of the future residual cash flows of thefirm but also to properly communicate all newsabout these cash flows to the financial markets.Only by disclosing this information to the financialmarkets potential information asymmetries can be

Figure 1. The neo-classical view on the objective of thecorporate firm.

W.G. HALLERBACH AND J. SPRONK188

Copyright # 2003 John Wiley & Sons, Ltd. J. Multi-Crit. Decis. Anal. 11: 187195 (2002)


3/9

resolved and the fair market value of a cash flowstream can be gauged. In evaluating the possibleconsequences of its decision alternatives, manage-ment should estimate the effects on the uncertain(future) cash flows followed by an estimationof the financial markets valuation of these effects.Then (and only then) the decision rule of manage-ment is very simple: choose the decision alter-native that generates the highest estimated market

value.The first problem that might arise while follow-

ing the above prescription is that residual claimscannot always be defined because of gamingeffects (see Figure 2, Box 2). In other words, thefuture cash flows of the firm do not only dependon the present and future decisions of the firmsmanagement, but also on the present and futuredecisions of other parties. An obvious example is

the situation of oligopolistic markets in which thedecisions of the competitors may strongly influ-ence each other. Similar situations may arise withother external stakeholders such as powerfulclients, powerful suppliers and powerful financiers.Games may also arise within the firm, for instancebetween management and certain key productionfactors. The problem with game situations is thatthe effect (on the firms future cash flows) caused

by the other parties involved cannot be treated assimple constraints or as cost factors in the cashflow calculation. MCDM may help to solve thisproblem by formulating multidimensional profilesof the consequences of the firms decision alter-natives. In these profiles also the effects on otherparties than the firm itself is included. Thesemultidimensional profiles are the keys to open thecomplete MCDM toolbox.

Figure 2. Situations leading to MCDM in the corporate firm.

RELEVANCE OF MCDM FOR FINANCIAL DECISIONS 189



4/9

The second problem in dealing with the singleobjective wealth maximization problem is that thequality of information concerning the firms futurecash flows under different decision alternatives isfar from complete (complete in terms of prob-

ability distributions and/or stochastic processes).In addition, the available information may bebiased or flawed. One way to approach theincomplete information problem is suggested bySpronk and Hallerbach (1997). In their multi-factorial approach, different sources of uncertaintyshould be identified after which the exposures ofthe cash flows to these risk sources are estimated.The estimated exposures can next be included in amulti-criteria decision method. In the case that theavailable information is not conclusive, differentviews on the future cash flows may developed.Next each of these views can be adopted as

representing a different dimension of the decisionproblem. The resulting multidimensional decisionproblem can then be handled by using MCDM(see Figure 2, Box 2).

The third potential problem in wealth maximi-zation is that the financial markets do not alwaysprovide the relevant pricing signals to evaluate thewealth effects of the firms decisions, for examplebecause of market inefficiencies. This means thatthe firm may want to include attributes otherthan the markets pricing signals alone in orderto measure the riskiness and wealth effects of itsdecisions.

3. MULTIPLE CRITERIA FINANCIALDECISIONS AND THE INVESTOR

Maximizing market value is the central paradigmin financial economic decision-making. There isone complication, however. Current market valueoriginates from future cash flows. A commonshare of stock, for example, derives its value fromthe dividends that are expected to be received inthe future (and from its subsequent selling price).

Likewise, the value of a bond depends on thecoupon payments and final repayment of thenominal face value. In a frictionless market(in which one can trade immediately at notransaction costs) one would indeed be indifferentbetween receiving $100 in cash, $100 in stocks or$100 in long term bonds. After all, the stocks orbonds can be sold immediately in the marketresulting in a cash position of $100.

The choice situation is completely differentwhen we rephrase the problem as holding aposition of $100 in cash, stocks or bonds over aperiod of say one year. Now we are confrontedwith the cash flows that will be generated by the

position during this holding period of one year,and with the value that will be realized at the endof the period. The degree of uncertainty attachedto the cash flows from a risk free position, a longterm bond and a stock is fundamentally different.It is therefore not feasible to maximize the returnover the holding period, for the simple reason thatthe return is stochastic. Maximizing expectedfuture return does not make sense either since thiscriterion ignores risk (strictly speaking, withprobability one the realized return will deviatefrom expected return).

How does financial theory solve this problem?

In the neo-classical approach to finance it isassumed that an expected utility function, definedin terms of future wealth or holding period return,is maximized. The optimal decision rule is derivedfrom confronting the preferences of the investorwith the probability distribution of future invest-ment returns. This allowed the Nobel laureateMarkowitz (1952, 1959) to formalize the formerlyundefined notion of risk by equating risk with thevariability of returns in a portfolio context, and tooperationalize this risk concept by means of the(co-) variance or standard deviation. From thisemerges the concept of a mean-variance efficient

portfolio: given its level of risk its maximizesexpected return, and given its level of expectedreturn it minimizes risk. The trade-off betweenexpected return and standard deviation embeddedin his preference structure allows an investor tochoose a suitable portfolio from the mean-variance efficient (i.e. non-dominated) set. In thisway the mean-variance decision criterion is a two-parameter substitutive criterion (see e.g. Sinn(1983)).

Note that the adopted preference functional isdefined on single dimensional wealth or return.This relates to the left hand side of Figure 2, where

there is only a single objective. But what about thetwo dimensions: expected return and risk? Well,because of the assumed substitutability betweenrisk and expected return these two dimensions canbe collapsed into a single preference dimension.This is even more apparent when one is willing toassume that risk free investment opportunitiesexist. In that case the optimal portfolio is simplythe portfolio that maximizes the uni-dimensional




5/9

Sharpe (1966, 1994) ratio (the Sharpe ratio isdefined as the ratio of expected return in excess ofthe riskfree rate, and the standard deviation).

Obviously, the simplicity of the mean-variancedecision rule goes at a cost. In order to derive the

decision rule, quite restrictive assumptions have tobe made either with respect to the preferences ofthe decision maker (quadratic utility) or withrespect to the representation of choice alternatives(joint ellipticity, cf. Owen and Rabinovitch, 1983).This is important since the mean-variance decisionrule is truly conditional normative in the spirit ofKeynes (1891). The decision rule is normative andguides to optimal decisions, conditional on satisfy-ing the underlying assumptions.

Now we turn to investment practice. How likelywill the assumptions underlying the mean-varianceapproach be satisfied? Even within the context of a

preference functional defined in terms of singledimensional wealth, variance may be an inade-quate risk measure. After all, there is an importantdifference between variability per s!ee and risk,especially when return distributions are asym-metric. For example, downside variability isperceived as risk, and this may explain the obvioussuccess with which the financial community hasembraced the downside risk metric Value-at-Risk(Jorion, 2000). Upside variability, in contrast, isevaluated positively as potential.

In addition, a single dimensional risk measuremay not prove sufficient to adequately discrimi-

nate between investment alternatives. In financialpractice it is recognized that the potential futurevariability in the returns can be attributed to thevariability in several state variables or economicfactors, which are beyond the control of thedecision maker. Some examples are interest rates,inflation, economic growth, and even psychologi-cal factors (such as market sentiment). Each ofthese factors represents a dimension of theeconomic environment in which the securityreturns are generated. We can thus view thereturns as being generated by the factors. Con-versely, the stochastic outcomes are conditionedon

these factors. This notion can be formalized by afactor model (see for example Sharpeet al., 1999).The relationship between a securitys return andchanges in each of these factors is described by aresponse coefficient or factor sensitivity. By meansof the factor sensitivities, the joint distribution ofsecurity returns is linked to (i.e. conditioned on)the joint distribution of factor changes. In thisinterpretation, the sensitivity coefficients can serve

as risk measures. After all, each sensitivity reflectsthe degree of exposure to the corresponding factorrisk. Taken together, the factor sensitivities or factorexposures constitute the multidimensional riskexposure profile of a decision alternative. Although

the factor changes are exogenous, the decisionmaker can adjust the risk exposure profile: he canchoose how sensitive he wants to be to each of therisk factors. As the variability of returns is linkedto the variability in various identifiable economicvariables, investment risk becomes an intuitivelyappealing, multidimensional concept.

The use of factor models permits replacingreturn variance as a uni-dimensional risk measureby multi-dimensional risk measures. These mea-sures provide more insight in the nature of riskperceived by investors than a uni-dimensional,aggregate risk measure. In order to shape the risk

profile of an investment portfolio not only thetradeoffs among the exposures to the variousfactors must be evaluated, but also the tradeoffbetween expected return on the one hand anddifferent factor risks on the other. It may be clearthat forcing this decision problem into thestraitjacket of a single dimensional objectivefunction with policy constraints will result in avery poor description of the actual decisioncontext. Obviously this provides a fruitful areafor MCDM applications.

In our opinion the multifarious nature ofperceived risk is only the first argument for

applying MCDM. Experiences from practice showthat explicit return and risk attributes do not seemto capture all relevant information (cf. the earlyresearch by Baker and Haslem, 1974, for example).To fill this gap, additional attributes may beincorporated at the investors discretion. Theseindirectly return related attributes may be con-sidered of general relevance in practice, but mayalso be relevant because of idiosyncrasies in theinvestors personal decision context. In the lattercase, the incorporation of additional attributes canbe motivated from either the specific tastes anddesires (goals) of the investor, from specific

investment constraints he faces, or from distinctivecharacteristics of the investment alternatives. Forexample, because of the investors tax situation,the taxability of the portfolio components may bea relevant attribute. In terms of liquidity or theflexibility to revise the portfolios composition,the marketability of the component securities maybe relevant. Especially the Long Term CapitalManagement debacle in September 1998 revealed




6/9

the crucial importance of market liquidity. Be-cause of some method of performance measure-ment, the position with respect to some benchmarkportfolio may be relevant, and so on. In addition,the investor may adhere to the notion that not all

future events can be reduced to probabilitydistributions, not even when the latter are of asubjective nature. This also implies that attributesmay be considered in addition to explicit elementsof return and explicit components of risk. Wemust seriously consider the possibility that someof these other attributes act in fact as proxies for(components of) expected return and risk.

Many attributes are considered important, notonly from a practical point of view, but also froman academic point of view because they representanomalies. (An attribute is an anomaly withrespect to an asset pricing theory when that

attribute possesses power to explain cross-sec-tional variation in expected returns in addition tothe risk measures as specified by the pricing modelat hand. An attribute is an anomaly with respect tothe efficient market hypothesis when it can be usedto forecast future returns. For an overview werefer to Sharpe et al., 1999). For common stocks,firm size is a long-time notorious variable. Otherexamples are price ratios as indicators for funda-mental firm value, like earnings/price, book/price(book value of common equity per share dividedby market price per share), cash flow/price, sales/price and dividend/price. In the context of value

investing there is great renewed interest in theselong time familiar attributes (see for exampleFama and French, 1992, 1993).

In the view of (positive) financial theory, anattributes ability to contribute to the explanationof cross-sectional return differences appears to bea convincing criterion for the selection of relevantattributes. However, an attribute will only carry asignificant premium when it is priced in themarket. But a non-average investor can face a setof investment opportunities that is different fromthe market (i.e. the average investor). For exam-ple, his opportunity set can be restrained by

investment restrictions or by prohibitive transac-tions costs. Hence this investor is only interested inthe relevance of this attribute in his opportunityset. Furthermore, partly connected to the formerargument, the reward that an investor attaches tothe exposure to an attribute (a subjectivepremium) may well be different from the premiumthat the market as a whole attaches to thatattribute (the objective premium). The difference

between objective and subjective premia justreflects the differences in the preference structureof the investorvis "aa visthe market as a whole. Thisleads us back to the starting point that theselection of attributes depends on the personal

circumstances of the investor, as summarized in hisprofile. In brief, there exist many security attri-butes that are relevant in practice, despite thestylized neo-classical view of financial theory.

How can the decision making process be shapedin this general multi-attribute context? Supposethat an investor can demarcate a set of securityattributes that he considers relevant. On the basisof these attributes he can discriminate between theattractiveness of various securities (this view isborrowed from consumer choice theory; cf. Lan-caster, 1966). For the investor, a financial securitythen represents a basket of, say, k attributes and

can be fully characterized by a k-tuple of attributevalues. In an investors view, when buying asecurity, he is actually buying an exposure tovarious attributes. Hence, we can specify amapping of the securities in the space spanned bythe attributes:

security i! ai1; ai2;. . .; aij;. . . ; aik

; i2 N 1

where aij is the value that attribute j takes forsecurity i. The opportunity set an investor facescan then be described in terms of a multi-attributerepresentation of the N available securities. For agiven portfolio, its score on a certain attribute is a

function of the attribute scores of the individualsecurities contained in this portfolio. The fractionsinvested in each of these securities can thus betreated as instrumental variables and the attributeexposures can be seen as goal variables. Theinvestment problem now becomes a multiattributeportfolio selection problem where the investorstrives to balance the attributes of the individualsecurities on the portfolio level. Given the securityattributes and the specific decision context, theinvestor strives to fashion the attributes of hisportfolio in a way that suits his particularcircumstances and preferences best. Often, the

investor will try to either minimize or maximizeeach of these goal variables. Alternatively, theinvestor may strive to attain a target level ordesired score on some attribute(s). Depending onthe investors insights and preferences, the relativeimportance of each of these goals may vary.Generally, no portfolio can be found for whicheach of the goal variables reaches its optimal valueor for which all criteria are met. As a consequence,




7/9

the investor has to evaluate the trade-offs betweenthe various goal variables. This calls for multi-criteria decision methods. In this context, a singleobjective decision framework would become aProcrustes bed, chopping off any other character-

istics of the investment alternatives that theinvestor may consider important.

The notion that the portfolio investment decisionfor individual investors calls for a multiattributeapproach dates back to Smith (1974, p. 53); Spronkand Hallerbach (1997) present a general frameworkfor moulding the financial investment decisionprocess, labelled multiattribute approach to port-folio selection. Of course, there are many moreexamples of moulding the portfolio selectionproblem in terms of multi-criteria decision making.However, one can observe that most (if not all) ofthese approaches are at best only partial. Either

they fail to give room for the inherent complexityof the decision procedure, or they concentrate onthe beauties of a particular multiple criteria deci-sion method taking insufficient account of thedecision context and of the results and principlesof financial economic theory.

In our opinion, portfolio problems signify afertile ground for applying MCDM methods.However, the potential rewards will only be reapedwhen the stage of applying MCDM methods to agiven financial decision problem is surpassed andMCDM and financial knowledge is integrated. Animportant additional desirable feature of the

applied MCDM technology is the possibility toexplore the decision context. Instead of hiding thesolution process and presenting the final solutionin one step to the decision maker, one should allowthe decision maker to learn about his preferencesand about the characteristics of the investmentopportunities. For example, the selection ofrelevant attributes is no once and for all activity.The investors decision context and the securitieseconomic environment may change over time andmay become better understood because of learningeffects. In addition the investor can evaluate thetrade-offs offered by the investment alternatives

and compare them with the desired trade-offs.

4. RISK MANAGEMENT

Financial markets are forward-looking: currentmarket value incorporates the markets expecta-tions regarding the future. Changes in theseexpectations will induce changes in value. This

implies that it is not only important to considercurrent market value as an important (but not thesingle) aspect of financial decisions, but also toincorporate the notion of potential future changesin this value. This calls for a risk analysis of the

decision alternatives, in which the sources of riskand the corresponding exposures are identified }and possibly quantified. Moreover, when timepasses by the current value must be guarded andprotected against unwanted influences that maycorrode this value. This in turn implies the needfor risk management: the process of adjusting riskexposures in such a way that the desired riskprofile is attained.

From the previous discussion, the role of riskanalysis and risk management for financial invest-ment decisions may be clear. Indeed, consideringthe level of volatility on financial markets the

claim that financial investment is mainly riskmanagement is not exaggerated. Especially in anasset-liability context (for banks, pension fundsand insurance companies) investment decisions aredriven by the desire to match the risk profiles ofthe liability portfolio on the one hand and theinvestment portfolio on the other. There aremultiple risk sources (interest rates, economicgrowth, inflation, e.g.) and choosing the desiredrisk profile entails evaluating the trade-offs amongthe different risk exposures and between the riskson one side and expected return on the other. Thiscalls for multi-criteria decision methods.

Even in the simple case where an investmentportfolio seems to be only sensitive to interest raterisk (a fixed income portfolio) a multi-criteriaapproach can add value. Since long, a bondsinterest rate exposure is measured by duration,defined as the (negative of the) bonds interest rateelasticity (see Fabozzi, 1997). However, the con-cept of duration assumes that the term structure ofinterest rates will only move in a parallel way. Aricher description of interest rate risk can beobtained by allowing any shift in the termstructure of interest rates. Instead of one singleduration measure one could use a set of partial

durations, each measuring the sensitivity of thebond price to changes in some specified rate. Thisis analogous to the multifactor model discussedbefore. Reitano (1990, 1996) and Ho (1992) specifya set of key spot interest rates and consider thebond price sensitivities for changes in these keyrates. Key rate durations may be used to accountfor the three major interest rate risk factors asidentified by Litterman and Scheinkman (1991):




8/9

changes in the level, slope and curvature of thespot rate curve. Positioning a fixed incomeportfolio in an uncertain interest rate environmentinvolves gauging the exposures to different sourcesof interest rate risk and evaluating the trade-off

among different exposures as well as between risksand returns. Especially when the trade-offs arenon-linear, the multi-criteria specification of thedecision problem cannot simply be collapsed into atwo-parameter risk-return substitutive criterion.When translating additional criteria into restric-tions, sight on these trade-offs is lost.

Risk analysis and risk management also play aparamount role in capital investment decisions. Inthe capital budgeting process the risk profile of thefuture cash flows generated by investment projectsmust be clearly understood. With the increasingglobalization of markets, increased competition

and fast changing consumer preferences, theeconomic environment in which these cash flowsare generated can change drastically in manydimensions. The firm then faces the difficult taskto evaluate a projects risk profile and } ifnecessary } to change this profile by means ofoperational or financial hedging techniques; cf.Smithson (1998) or Stulz (2003). Crucial determi-nant of the decision to hedge is the degree ofcomparative advantage that the firm has in thearea of a specific risk source. When the compara-tive advantage is high (as will be the case for thefirms core activities) the firm can add value for its

shareholders (and other stakeholders) by main-taining an exposure to the corresponding source ofrisk. Conversely, the firm may decide to hedge riskexposures that are not (or only remotely) related toits core activities. In any case, the decision tohedge involves evaluating the trade-offs betweenthe costs and benefits attached to exchanging oneexposure for another. The interdependency betweenrisk sources makes it difficult to evaluate thesetrade-offs. This is where multi-criteria methodsmay help to support the firm in its hedge decisions.

Consider for example the airline KLM. KLMexperiences the influences of different risk sources:

it has exposures to foreign currencies (especiallythe US dollar with respect to the euro), interestrates (mainly because of its financing structure)and jet fuel prices. The hedging decision involvesevaluating the risks and potential returns fromthese various risk sources. However, this riskanalysis cannot be performed only for KLM inisolation. Also the potential effects of competitorshedging strategies on KLMs position have to be

taken into account (see Figure 2). Suppose, forexample, that competitors hedge but KLM doesnot. When jet fuel (or oil) prices increase, KLMexperiences a disadvantage relative to its compe-titors and this may hurt its profitability. When

KLM would increase fare prices to compensate forthe increased cost, its relative competitive positionmay deteriorate, resulting in loss of market share.Of course, KLM would experience a relativeadvantage when jet fuel prices decrease but theeffect is not symmetrical. After all, it is much moredifficult to conquer market share than to lose it.

5. CONCLUSION

From the neo-classical perspective of perfectmarkets, financial decision problems can be

moulded in terms of a single objective. However,imperfections such as information asymmetries,conflicting interests and transactions costs (whichrestrict choice opportunities) require a much richerdescription of the decision context.

Focusing on the three main areas of finance }corporate finance, financial investment and riskmanagement } we argue that many decisionproblems involve multiple objectives, and some-times even multiple actors. This leads to decisionproblems with multiple criteria. The fundamentalconcept of risk, for example, is by nature multi-dimensional. Competitive forces to the corporate

firm are partly driven by decisions of third parties.Consequently, a single objective approach tofinancial decision making becomes a straitjacket.When attempting to cope with additional objec-tives by rephrasing them as policy constraints, thesingle objective feature is preserved. However, thisgoes at the high cost of losing the informationabout the relevant tradeoffs.

The multidimensional nature of many financial-economic decision problems lends itself for theapplication of MCDM technology. However, toensure a suitable description of a particular deci-sion context and to choose a suitable MCDM

method to solve the decision problem at hand,financial theory and MCDM should be broughtcloser together. Problem description and problemsolving are not two separate stages towards makingbetter decisions. Instead, the whole trajectory fromanalysing the decision context and gathering(imperfect and incomplete) information to choos-ing and implementing a suitable MCDM technol-ogy should be spanned. Indeed, the complexity of




9/9

financial choice problems calls for an integrationof financial and MCDM knowledge. Not only toimprove final decisions, but above all to improvethe quality of the choice process. The latter isnecessary in order to cope with the frequent

changes in the decision context. It allows benefit-ing from learning effects and it may at the sametime facilitate the wider acceptance of MCDMtechniques in practical decision making.

REFERENCES

Baker HK, Haslem JA. 1974. Toward the developmentof client-specified valuation models. The Journal ofFinance 29(4): 12551263.

Fabozzi FJ. (ed.). 1997.The Handbook of Fixed IncomeSecurities. Irwin: Chicago, Ill.

Fama EF, French KR. 1992. The cross-section of expected

stock returns. The Journal of Finance 47: 427465.Fama EF, French KR. 1993. Common risk factors in

the returns on stocks and bonds. Journal of FinancialEconomics 33: 356.

Ho TSY. 1992. Key rate durations: measures of interestrate risks. The Journal of Fixed Income 2(1): 2944.

Hull JC. 1999. Options, Futures and Other Derivatives.Prentice-Hall: Upper Saddle River, NJ.

Jensen MC. 2001. Value maximization, stakeholdertheory, and the corporate objective function.Journalof Applied Corporate Finance 14(3): 821.

Jensen MC, Meckling WH. 1976. Theory of the firm:managerial behavior, agency costs and ownershipstructure. Journal of Financial Economics 3: 305360.

Jorion Ph. 2000. Value at Risk: The Benchmark forControlling Market Risk. McGraw-Hill: Chicago, Ill.

Keynes JN. 1891. The Scope and Method of PoliticalEconomy.Macmillan: London.

Lancaster KJ. 1966. A new approach to consumertheory.Journal of Political Economy 74: 132157.

Litterman R, Scheinkman J. 1991. Common factorsaffecting bond returns. The Journal of Fixed Income1: 5461.

Markowitz HM. 1952. Portfolio selection.The Journalof Finance 7(1): 7791.

Markowitz HM. 1959. Portfolio Selection: EfficientDiversification of Investments. John Wiley: New York.

Martel JM, Zopounidis C. 1998. Crit "eeres multiples etd!eecisions financiers, Revue FINECO 8: 1,2.

Owen J, Rabinovitch R. 1983. On the class of ellipticaldistributions and their applications to the theoryof portfolio choice. The Journal of Finance 38(3):745752.

Reitano RR. 1990. Non-Parallel yield curve shifts anddurational leverage.The Journal of Portfolio Manage-ment 16(4): 6267.

Reitano RR. 1996. Non-Parallel yield curve shifts andstochastic immunization. The Journal of PortfolioManagement 22(2): 7178.

Sharpe WF. 1966. Mutual fund performance.Journal ofBusiness 39(1): 119138.

Sharpe WF. 1994. The sharpe ratio.Journal of PortfolioManagement 21(1): 4958.

Sharpe WF, Alexander GJ, Bailey JV. 1999.Investments.Prentice Hall: Upper Saddle River NJ.

Sinn HW. 1983. Economic Decisions Under Uncertainty.North-Holland: Amsterdam.

Smith KV. 1974. The major asset mix problem of theindividual investor.Journal of Contemporary Business,Winter: 4962.

Smithson CW. 1998.Managing Financial Risk: a Guideto Derivative Products, Financial Engineering, andValue Maximization. McGraw-Hill: New York.

Spronk J, Hallerbach WG. 1997. Financial modelling:where to go? With an illustration for portfoliomanagement. European Journal of Operational Re-search 99: 113125.

Steuer RE, Na P. 2003. Multiple criteria decisionmaking combined with finance: a categorized biblio-graphy. Special issue on financial modeling.EuropeanJournal of Operational Research, forthcoming.

Stulz RM. 2003. Risk Management and Derivatives.Thomson South-Western: Mason Ohio.

Zopounidis C. 1999. Multicriteria decision aid infinancial management. European Journal of Opera-tional Research 119: 404415.



Journal of Multi-Criteria Decision Analysis Volume 11 issue 4-5 2002 [doi 10.1002_mcda.328] Winfried...

Documents

Transcript of Journal of Multi-Criteria Decision Analysis Volume 11 issue 4-5 2002 [doi 10.1002_mcda.328] Winfried...