Handbook of Systems Analysis: Volume 1. Overview. Chapter 6. Objectives, Constraints, and Alternatives

(1)

NOT FOR QUOTATION WITHOUT P E R M I S S I O N O F THE AUTHOR

HANDBOOK O F SYSTEMS A N A L Y S I S VOLUME 1 . OVERVIEW

CHAPTER 6 . O B J E C T I V E S , C O N S T R A I N T S , AND ALTERNATIVES

E d w a r d S . Q u a d e

O c t o b e r 1 9 8 1 W P - 8 1 - 1 4 2

V o r k i n g P a p e r s a r e i n t e r i m r e p o r t s on w o r k of t h e I n t e r n a t i o n a l I n s t i t u t e f o r A p p l i e d S y s t e m s A n a l y s i s a n d have received o n l y l i m i t e d r e v i e w . V i e w s o r o p i n i o n s e x p r e s s e d h e r e i n do

n o t

n e c e s s a r i l y repre- s e n t t h o s e of t h e I n s t i t u t e

o r

of i t s N a t i o n a l M e m b e r O r g a n i z a t i o n s .

INTERNATIONAL I N S T I T U T E FOR A P P L I E D SYSTEMS A N A L Y S I S A - 2 3 6 1 L a x e n b u r g , A u s t r i a

(2)

FOREWORD

The International Institute for Applied Systems Analysis is preparing a Handbook of Systems Analysis, which will appear in three volumes:

Volume

1 :

Overview is aimed at a widely varied audience of producers and users of systems analysis studies.

Volume

2:

Methods is aimed at systems analysts and other members of systems analysis teams who need basic knowledge of methods in which they are not expert; this volume contains

introductory overviews of such methods.

Volume 3: Cases contains descriptions of actual systems analyses that illustrate the diversity of the contexts and methods of systems analysis.

Drafts of the material for Volume 1 are being widely

circulated for comment and suggested improvement. This Working Paper is the current draft of Chapter 6. Correspondence is invited .

Volume 1 will consist of the following ten chapters:

1. The context, nature, and use of systems analysis 2. The genesis of applied systems analysis

3. Examples of applied systems analysis

4. The methods of applied systems analysis: An introduction and overview

5. Formulating problems for systems analysis 6. Objectives, constraints, and alternatives

7. Predicting the consequences: Models and modeling

. 8.

Guidance for decision 9. Implementation

10. The practice of applied systems analysis

To these ten chapters will be added a glossary of systems analysis terms and a bibliography of basic works in the field.

12 October 1981

Hugh

J.

Miser IIASA

A-2361 Laxenburg

Austria

(3)

CONTENTS

1

.

O B J E C T I V E S

. . .

¹

. . .

Measuring effectiveness 8

Multiple objectives

. . .

^{1 1}

2

.

CONSTRAINTS

. . .

^{1 4}

. . .

3 . A L T E R N A T I V E S . 1 7

REFERENCES

. . .

^{2 3}

(4)

CHAPTER 6. OBJECTTVES. CONSTRAINTS. AND ALTERNATTVES

Edward. S. Quade

To help a decisionrnaker means to help hlm a c h e v e h s t r u e objectives; to do so, it is crucial to discover what they a r e . To find feasible alternatives (l.e., ways Lo a c h e v e the objectives that are not ruled out by the constraints imposed on the decisionmaker's actions by n a t u r e , circumstance, authority, or the decisionmaker himself) is the central task of systems analysis. T h s chapter, expanding the remarks in Chapter 4, discusses problems of clarification and measurement related t o objectives, t h e roles of objectives and constraints in determining alternatives, and methods for discovering and improving alternatives, and eliminating t h e inferior ones.

1. OBJECTIVES

Ideally, problems with objectives (and witn certain constraints, since they play a role sonlewhat similar to t h a t of objectives) should be settled very early in a systems analysis study. In practice, however, this is seldom possible.

Unfortunately, a clear, well thought through, precisely spelled out, and analyti- cally useful statement of what the decisionmaker wants to accomplish is rarely presented to the analyst a t the tlme t h e study is commissioned. Kevertheless, some idea of what is wanted is available. Tentative objectives, specific enough to get the analysis s t a r t e d , can b e selected; these should suggest possible ways to

(5)

a c h e v e the alternatives. The impacts, or consequences, of irriplementing these alternatives a r e then imagined or estimated, and their projected implications used to reexamine t h e first formulation of objectives and introduce modifications. As t h e study progresses, the analysts, their sponsors, and others learn from t h s early work. The decisionmaker for whom the work is being carried out is also influenced by pressure from interested constituents and from other decisionmakers who may see their domain adversely affected by what they anti- cipate is likely t o be done as a result of the study. Hence objectives and alternatives may change and constraints may be introduced or removed. This is one of the major reasons why systems analysis must be a n iterative process.

To someone without systems analysis experience it may s e e m odd t h a t we s t a r t this discussion with t h e implied assumption that t h e objectives will not be spelled out sufficiently t o allow the analysis team to adopt them a s given and proceed to their work. However, experience is almost universal on this point:

decisionmakers seldom have carefully articulated objectives. T h s apparently troublesome fact-whch leads t o t h e iterative s e a r c h for realistic and acceptable objectives as part of t h e systems analysis-may lead the wishful analyst into dreaming of the "perfect client" who is the exception, so t h a t he c a n get quickly beyond the beginning s t a t e m e n t of objectives t o what he may think of as the meat of the analysis. However, here again the voice of experience suggests t h a t it may be b e t t e r not t o have such clearly defined obejctives right a t t h e beginning before a t least some preliminary analysis has taken place, for t h e objectives stated a t t h e beginning will probably not be a s well thought through a s they should b e , with the result t h a t the analyst will face the dilemma of tailoring his work to objectives he knows t o be defective, or trying a t the end to alter these fixtures in order to squeeze in some new perspectives. The experienced analyst does not rue the existence of hazy starting objectives; r a t h e r , he is thankful f o r t h e m , a s they offer h m the opportunity t o work with his client toward realistic and acceptable ones a s t h e study proceeds.

(6)

4 '

The objectives of an individual o r of an organization a r e t h e principles t h a t d e t e r m i n e how they a r e to a c t . As Sugden and Williams (1978) p u t it, "They a r e not pious incantations of ethical or ideological beliefs which unfortunately cannot be a c t e d on because of 'political c o n s t r a i n t s ' ; they e x p r e s s intentions t o a c t . " The analyst needs s o m e t h n g s p e c i f ~ c enough t o limit t h e s e t of alternatives and t o guide t h e choice among t h e m . In p a r t i c u l a r , h e would like this specificity t o be such t h a t he c a n tell t h e decisionmaker ( a s soon as he c a n get s o m e clue a s to what they a r e ) what s o r t s of alternatives a n d consequences a r e likely t o result. The decisionmaker c a n t h e n conlirm t h a t t h e s t a t e m e n t of objectives expresses his intentions and offer a n o t h e r suggestion if it does n o t .

Even when objectives a p p e a r t o be well specified a t t h e o u t s e t , they c a n sel- d a m b e adopted uncritically by the analyst. Means a r e s o m e t i m e s t a k e n for e n d s ; a decisionmaker m a y say t h a t h s objective is t o find out where t o place a new comprehensive medical c e n t e r in his district, b u t his real objective m a y be b r o a d e r , say to improve t h e totality of health servlces in hls community; b e t t e r ways t o achieve the real objective m a y be t o provide several neighbornhood health c e n t e r s or services t h r o u g h o t h e r m e c h a n i s m s (hospital outpatient clin- ics o r health maintenance organizations). P e r h a p s p r o g r a m s focusing on m a t e r - nal a n d c h l d health services or screenin.g a p p a r e n t l y well people t o t u r n up h e a r t and c e n c e r conditions should be considered. Unless t h e b r o a d e r objective is investigated by t h e analyst, t h e s e l a t t e r alternatives will not a p p e a r and t h e decisionmaker may never realize ho-w m u c h m o r e h e cou.ld do with his resources.

Desired goals c a n be s t a t e d easily, b u t specifying objectives one has a r e a - sonable hope of attaining m a y t a k e considerable t h o u g h t , even r e s e a r c h . It m.ay not m a k e sense t o m a k e up o n e ' s mind as t o what one wants until one has a fair idea as t o what h e c a n g e t . An i m p o r t a n t way analysis helps in clarifying objectives is t h a t it d e t e r m i n e s t h e undesi.rable a s well a s t h e desirable consuqences of t h e altc!rn.atives t h a t follow a n assumpti011 a b o u t objectives. The decision- m a k e r , when confronted with t h e s e consequences, c a n ask himself whether he is

(7)

willing to accept what they imply. If not, he will have t o modify his goals.

In the effort t o clarify objectives and find ways to measure their attain- m e n t , it is helpful to discuss the issue under analysis with several people who have no stake in t h e outcome-critical and skeptical outsiders. One seeks answers to questions such as "What is really a t stake here?" Since attainment of the ultimate goal may be many years off, what practical intermediate goals should we strive for?

For certain issues, t h e question of whose objectives a r e relevant h a s t o be considered. For public issues, it is some subset of t h e citizens of today o r of future generations; the decisionmaker is merely t h e person o r organization charged with the responsibility for changing the system, and t h e analyst may have to find some discrete way t o make t h s clear (Sugden and Williams 1978, pp. 232-242).

High-level objectives tend t o be ambiguous, for it is a political advantage t o appeal to as many people as possible. In fact, it is often m u c h easier in the political process to agree on an action or program t h a n on a goal (unless t h e l a t t e r is indefinite), for people may have different motives for what they are willing t o do. High-level objectives t h a t express general good intentions a r e valid over a long time period and a r e the easiest for the decisionmaker t o s t a t e (Hovey 1968). A frequent problem is t h a t such objectives are difficult for the analyst t.o formulate with sufficient specificity for direct use in analytical studies. A lower- level objective t h a t is a means to achieve the hlgher-level objective may be required for analysis. To build a new health center is both a lower-level objective and a means t o improve health services, a higher-level objective. Clear definitions of lower-level objectives a r e usually more easily provided and a r e techni- cally easier to use for finding and ranking alternatives. However, misleading results may occur if t h e lower-level objectives are not, under all cir.cumstances, an appropriate mehns t o achieve the higher-level objectives. The relation need not be direct; for example, t o relocate fire stations t o provide better fire service, the objective of lowering t h e average travel time from fire station t o fire

(8)

serves well as a substitute for b e t t e r protection of lives and property (Chapter 3 and Walker e t al. :979). Travel time is, moreover, easily measurable and can thus serve a s an estimate of the extent t o which the alternatives attain t h e objective.

Capabilities designed with one objective in mind can differ considerably from those designed with another uppermost, even though both have the same higher-level goal. T h s is a difficulty that c a n arise when one a t t e m p t s t o make a relatively vague higher-level objective operational by using a m o r e specific lower-level formulation. Wohlstetter (1964) illustrates t h s point for the case of automobile traffic safety:

The city fathers would like t o reduce the number of violations of the law. They would also like to fine o r put in the clink as many viola- t o r s as they can. There a r e two well known alternative techniques for accomplishing these ends; one is t h e familiar ambush technique; t h e other is sometimes called t h e visible patrol technique. The first increases the probability of interception and a r r e s t . The second discourages culpability. Now if our goal is to maximize the number or proportion of speeders punished, or the total of municipal revenue through fines, probably t h e best way to do the job is by ambush, however uneasy such a sneaky tactic makes us. If our goal, on t h e other hand, is a reduction in t h e total number of traffic accidents, say, or in t h e number of a t t e m p t s t o violate the law (even if on the whole such a t t e m p t s a t evasion as do take place a r e more likely to be successful, since the culprit is aware of t h e cop's presence), it may very well be that the most frequent, obvious presence of policemen capable of instantaneous retaliation against speeders would encourage caution, and so achieve such goals best.

I t is interesting to review t h e cases discussed in Chapter 3 in t.he llght of these r e m a r k s about objectives:

(9)

The basic objective of t h e Greater New York Blood P r o g r a m is t o have blood available when and where needed-always. The only admissible exception is t h a t elective uses c a n possibly, when necessary, b e postponed, although t h e costs of s u c h decisions a r e recognized a s making t h e m undesirable. Deriving from this basic objective-whch was b e i ~ m e t adequately when t h e systems analysis was undertaken-are two subordinate objectives: t o reduce t h e a m o u n t of blood being outdated ( a n d h e n c e discarded) t o the lowest level consistent with t h e basic objective, and t o r e d u c e t h e costs of maintaining t h e blood supply a t t h e points of use a s m u c h a s possible consistent with t h e basic objective. The simplicity and clarity of t h e s e objectives contributed a g r e a t deal to t h e crisp clarity of t h e analysis a n d t h e success of its implementation, because t h e s e objectives were widely a c c e p t e d .

The basic objective of t h e Wilmington Bureau of Fire is t h a t of any fire d e p a r t m e n t : t o hold down a s m u c h a s possible t h e h u m a n ( d e a t h a n d injury) and property losses from fire. However, owing t o t h e lack of information in t h e fi.re-fighting community a b o u t t h e relations between t h s objective and fire- fighting deployments a n d activities, t h s Bureau-and t h e systems analysts called in t o help with its planning problem-accepted t h r e e objectives t h a t were proxies for t h e basic objective: t o k e e p travel times t o individual locations down t o a n acceptably low level, t o k e e p t h e average travel t i m e in a region a t a n acceptable level, a n d t o maintain reasonable workloads for fire companies. As in t h e c a s e of t h e Greater New York Blood P r o g r a m , before t h e s y s t e m s analysis began t h e Wilmington fire fighters were a c h e v i n g t h e i r basic objective t o a level acceptable t o the community, s o t h a t one could argue pragmatically t h a t t h e existing level m e t t h e basic objective, and t h e words "acceptable" and "reasonable" in the s t a t e n - e n t of t h e proxy objectives imply t h a t fire-fighting officials will exercise judgments b a s e d o n t h e c r a f t skills t h a t t h e y have l e a r n e d f r o m m a n y y e a r s of experience. Thus, t h e objective of t h e analysis could be t o achieve a redeployment p a t t e r n maintaining t h e level of a c h e v e r n e n t of t h e basic objective, possibly achieving economies, o r raising t h e s t a n d a r d of service

(10)

as judged by the proxy objectives

In the case of the study of the Oosterschelde estuary, the basic objective was to protect it irorn future flooding caused by violent storms, but the level of protection to be provided, as well as t h e myriad other objectives (ranging from the concerns about microscopic organisms t o those of humans) were all open for analysis. While the basic objective was widely agreed t o , the analysts could count on no such consensus for t h e others, and, i n fact, had t o keep them all flexibly in view, in order t o allow the many parties of concern t o make their own judgments about them.

In the case of the IIASA study of t h e world's energy future for the next fifty years, since there was no world energy executive-or even a closely knit set of officials who could be thought of as an approximation-the analysts had to attribute objectives t o the world that would seem reasonable to the many execu- tlves and analysts who might later want to use t h e study's findings as a contribu- tion to their own thinking about their narrower problems. In broad t e r m s t h e basic objective assumed by t h e analysts was t h a t the world would want to sustain its use of energy indefinitely a t a level that would maintain t h e standard of living in developed parts of the world and allow the developing p a r t s t o increase their standard, whle accepting the population growth t h a t is expected. Supporting this basic objective there were many other objectives-for example, making the world's energy future grow by a natural and fairly smo0t.h evolution from the present world energy posture. While most of these supporting objectives were expected t o be widely subscribed to, variant views were expected on many-an expectation that proved to be realistic. For example, some solar energy enthusiasts feel t h a t the assumption of a sustained standard of living is inap- propriate, that a smooth transition should be replaced by a sudden switch to solar energy, and that future world energy use should be assumed to undergo a significant per capita reduction.

In sum, it is important to consider and evolve a clear concept of the objectives, and to seek as much of a consensus on them as possible. However, our

(11)

examples suggest t h a t the consensus may not be achievable, particularly a t the beginning of the analysis, when the consequences of various objectives may not be perceived realistically by all concerned. On t h e other hand, when there are myriad objectives, many conflicting-as in the Oosterschelde analysis-the work can illuminate them and their consequences so as to allow accommodation among potentially conflicting parties in the process of deciding on a course of action. All of these m a t t e r s emphasize t h e importance of having a clear view of objectives, all of t h e m that a r e relevant, even if they cannot be rationalized into a congenial s e t .

In brief, finding t h e right objective is crucial, more important than it is to find the very best alternative. The wrong objective means the research is devoted to the wrong problem; to designate a slightly inferior alternative is not nearly so serious.

Measuring effectiveness. To be considered an alternative, a course of action must be consis tent with the decisionmaker's objectives, i.e ., offer some hope of attaining the objectives, or a t least of bringing them closer. For the pur- pose of quantitative analysis we would like a scale on which the effectivenesses of the various alternatives in attaining the objectives can be measured. The alternatives can then be compared and ranked in terms of effectiveness, that is, by their standings on the effectiveness scale.

Unfortunately, a satisfactory scale cannot always be found easily. The problem is that many objectives are difficult o r impossible t o quantify directly in any useful f a s h o n . It is therefore necessary to use a surrogate or proxy, a substitute objective that can be measured and that approximates the extent t o w h c h the real objective is attained. The problem is to g e t a good approximation.

Thus, to measure the quality of medical care in a community, the infant mortal- ity r a t e is often used as a proxy, even though it merely measures one aspect of the quality of medical c a r e ; to measure fire-department performance, response time is often used (Walker e t al. 1979, p . 81).

(12)

One technique f o r flnding quantifiable ways t o m e a s u r e effectiveness is t o t r y several successive modifications of tentative possibilities for stating t h e objective t o s e e whether any substitutes or proxies a r e s u g g e s t e d . Sometimes it c a n be e a s i e r to examine several possible proxies to s e e w h e t h e r , if they were a t t a i n e d , t h e desired e n d would also be achieved.

The m a r k of a good proxy is t h a t its a c h e v e m e n t closely reflects achieving the real objective. Unfortunately, t h e r e a r e a n u m b e r of m a d e q u a t e b u t common a p p r o a c h e s in use.

One is t o use input t o m e a s u r e output; t o c o m p a r e t h e quality of primary school education in various districts in t e r m s of e x p e n d i t u r e s p e r pupil. A second is t o use workload m e a s u r e s or efficiency m e a s u r e s t o c o m p a r e quality of o u t p u t , s a y , t o c o m p a r e t h e quality of p r i m a r y e d u c a t i o n on t h e basis of teacher-pupil ratios.

Consider a single unambiguous objective, say, t o improve garbage collection. To facilitate comparisons, it is useful t o have a scale on which t o m e a s u r e the effectiveness of t h e various possibilities. But t h e r e is no obvious scale t o m e a s u r e b e t t e r g a r b a g e collect,ion, so we need a proxy-a s u b s t i t u t e objective m e a s u r e d b y , s a p , t h e p e r c e n t a g e of cit.y blocks remaining without health hazards, o r t h e r e d u c t i o n In t h e n u m b e r of fires involving uncollected solid waste, o r r o d e n t bites, or valid citlzen complaints. All of t h e s e , unfortunately, t r e a t just a n a s p e c t , n o t t h e full value, of b e t t e r garbage collection. In p r a c t i c e , people often u s e even less satisfactory scales, for i n s t a n c e , a n input measure-expenditure p e r household-or a n efficiency m e a s u r e - n u m b e r of tons collected p e r man-hour-or a workload measure-tons of waste collected-that indicate nothing a b o u t t h e quality of t h e work.

When several a t t r i b u t e s need to be considered, a combination is sometimes used in which t h e various a t t r i b u t e s a r e assigned weights, resulti-ng in a n ordinal o r cardinal utility function. The failing here is t h a t t h e function is t o a large e x t e n t t h e p r o d u c t of t h e analyst's judgment a s t o t h e relative importance of t h e a t t r i b u t e s a n d n o t t h a t of t h e responsible decisionrnakers. The

(13)

decisionmakers, if they were willing t o spend the time, could work out their own s e t of weights (with guidance from the analyst) but even here the analyst's influ- ence is powerful. Hatry ( 1970) comments:

There is no doubt t h a t the job of decisionmakers would be easier if a single effectiveness measure could approximately be used. However, I contend t h a t s u c h procedures place the analyst in the position of making a considerable number of value judgments t h a t rightfully should be made in the political decision-makirg process, and not by the analyst. Such value judgments a r e buried in the pr.ocedures used by the analysts and a r e seldom revealed to, or understood by, the decision makers.

Such hocus pocus in the long r u n tends to discredit analysis and distract significantly from what should be its principal role: to present t o decision makers alternative ways of acheving objectives, and t o esti- m a t e and display all t h e major tradeoffs of cost and effectiveness t h a t exist among these alternatives. [Emphasis in the original.]

The Oosterschelde analysis described in Chapter 3 shows, not only t h a t i t is not necessary to push disparate measures of effectiveness into a common meas- u r e , but also displaying t h e m with their relevant objectives can have important values for t h e decisionmakers.

In selecting a scheme t o measure effectiveness, we a r e not only looking for a scale t h a t is positively correlated with t h e objective under consideration but also for one for which t h e required d a t a can be obtained.

Suppose, for example, a n ongoing government program for training com- p u t e r operators is to b e evaluated. If t h e object~.ve of t h e program is the even- tual increase in the gross domestic product (GDP), how c a n effectiven.ess be measured? One possibility is t o calculate the incr-ease in GDP caused by t h e increase in t h e income of the trainee that results from the training he receives;

this requires t h a t the in.come of the trainee be total compensation (i.e., include fringe benefits) and t h a t adjustment be made for displacing previous workers, if

(14)

any, by t h e trainees.

But how do we actually get t h e needed information? By following the his- tory of t h e trainees after they leave the program, it is possible t o estimate their actual income, b u t how much of any change can be attributed t o t h e training?

One way to g e t t h e desired information would be t o carry out a social experiment (Chapter 7) in which participants a r e assigned randomly either to the program or to a control group, and then follow the wage experience of both groups.

But the experiment could well be much more costly t h a n the training program alone. A practical substitute would be t o use t h e wage experience of a group of people having similar backgrounds to the trainees for comparison purposes. It is imperfect, b u t conceptually correct; a measure based on it is likely t o be superior t o a n output measure such as the percentage of trainees t o get jobs as computer operators.

Multiple objectives,A decisionmaker may have more t h a n one objective. If so, they may conflict; he may wish t o reduce expenses but increase staff or to increase highway safety on a motorway between two cities but to decrease the travel time. In any event, if t h e r e a r e two or more objectives, t h e y compete (except possibly for pairs where one is the means to the other) in t h e sense t h a t , for given resources, if the decisionmaker strives for maximunl attainment of a particular one, he must accept less than maximum attainment of the o t h e r s .

As is obvious, one cannot maximize benefits and mimmize cost simultane- ously or do s o m e t h n g similar for any other pair of goals. But if the measures of attainment for these goals have a common unit, one can c r e a t e a new goal to achieve the most advantageous combination. For instance, one can maximize benefits minus costs ( a s in cost-benefit analysis) provided both can be expressed in monetary units (Nagel and Keef i979).

If t h e r e a r e several decisionrnakers, each with his own s e t of objectives, a number of different approaches c a n be used to expedite the process of reducing the number of objectives t o something t h a t can be used as a basis for analysis (Eilon 3.972, Keeney and Raiffa 1976). The following list gives some

(15)

examples-each of these approaches requires discussion among the decisionmakers holding competing objectives and a certain amount of compromise and concession:

Objectives t h a t are only means to achieve other objectives can be eliminated.

If all t h e objectives can be interpreted as means to achieve some higher- level objective, and a relevant way to measure its effectiveness or t h a t of a good proxy corresponding to this objective can be found, t h e n this higher-level objective may serve a s the single objective.

A preference ordering of objectives can sometimes be s e t up and used t o effect tradeoffs among them. A solution is first determined using the highest- ranking objective; t h e n a n effort is made to a c h e v e a s much of the second as possible without sacrificing too much of the first; and so on.

All objectives except (the most important) one c a n be converted into constraints, by agreement on a minimum level of attainment acceptable on each.

Tradeoffs among t h e objectives can be worked out and used to construct a composite index of worth, a value or utility function (decision analysis; see Raiffa 1968).

No effort can be made t o "optimize" with respect to any specific objective. Instead, all objectives c a n be converted into constraints and a s o l u t i o ~ determined under t h e agreement t h a t any solution satisfying all constraints-called a satisficing solution-will be "good enough."

It is, of course, not always possible t o r e a c h t h e agreement necessary t o implement any of these sim.plifications, although the use of special techniques t o increase the value sensitivity of decisionmakers (Dror 1975, p. 250) may make them more amenable t o compromise. We will t r e a t value sensitivity in a section under that title in Chapter 8 after discussing satisficing and the more common schemes for presenting results: cost-effectiveness and cost-benefit analysis, decision analysis, and the so-called "scorecards."

(16)

What we cannot do satisfactorily is construct a unique g r o u p objective from all the individual objectives t h a t automatically weights all the s e p a r a t e ones.

Arrow ( 1 951) has proved, Tor example. t h a t , under a f e w reasonable assumptions, t h e r e is no general procedure for obtaining a group ranking from the rankings of the individual group m e m b e r s .

Some idea of the interplay among objectives and t h e i r m e a s u r e s can be gained from looking m o r e closely a t t h e blood availability and utilization example. The basic objective is, a s we said earlier, t o have blood available when and where needed. However, t o this baldly idealistic objective-which could well entail very h g h costs-was added a qualification: t h a t elective uses of blood be allowed to be postponed when supplies a r e s h o r t . Within t h e framework of 2. sys- t e m meeting this goal, t h e analysts considered two additional objectives: t o reduce the amount of blood being outdated (and hence discarded) and to reduce t h e costs of maintaining t h e blood supply a t the points of u s e ( t h e proxy meas- u r e for t h s objective being t h e average n u m b e r of weekly deliveries t o a hospital blood bank).

The basic objective could be achieved by maintai.ning very large supplies a t t h e hospital blood banks a t all times, but this policy would entail very large outdating r a t e s ; in f a c t , most a r e a s in the United S t a t e s were pursuing policies dominated by t h s s o r t of approach and were experiencing outdating r a t e s of 0.20-that is, a fifth of t h e blood donated was not put t o its ultimate good use.

On the o t h e r hand, very low outdating r a t e s could be achieved merely by keeping minimal stocks a t t h e hospital blood banks-but this would r u n t h e risk of prejudicing seriously t h e basic objective of availability where needed whenever needed. Another policy could possibly m e e t the demands of t h e basic objective and t h e desire t o r e d u c e outdating t o a. minimum: keep all blood centrally and deliver expeditiously when needed. Very little speculation is needed t o d e t e r - mine the unacceptability of this policy; in addition t o escalating delivery costs, t h e r e would be unacceptable time delays in t h e face of s h o r t - t e r m health e m e r - gencies.

(17)

The study then was aimed a t a compromise among all of these objectives, and a c h e v e d it by keeping t h e basic objective a t its former high and acceptable level while reducing the outdating and shortage r a t e s (these l a t t e r causing extra deliveries to be made from the c e n t e r to the hospital blood banks). The results of t h e first t e s t were: outdating reduced from 20 percent to 4 percent, average weekly deliveries per hospital blood bank reduced from 7.8 to 4 . 2 (these figures also being a n indication of the increased adequacy of the hospital blood bank's supply, especially since 1 . 4 of the G.2 weekly deliveries were to cover shortages, the other 2 . 8 being routine prescheduled deliveries). Some experimentation with their models convinced the analysts that the 4 percent outdate r a t e was about the smallest t h a t could be achieved in t h s balance among the three objectives; i n other words, this outdate r a t e represented a cost of pursuing a balance, r a t h e r than just t h s one objective.

Similar discussions could be generated for the other examples in Chapter 3, but they would of necessity be much longer-too long for the space available here. Nor w ~ u l d they contribute essentially new insights. However, the diligent r e a d e r would find the sources from which we have drawn our brief accounts a rich mine of instructive issues of the sort indicated in the blood-bank discussion.

The objectives suggest the alternatives, the constraints restrict them and reduce the number of possibilities that can be considered. Constraints a r e thus more likely t o simplify than t o complicate the work of the analyst.

Constraints often have a function similar to objectives from a n evaluation point of view. Herbert Simon (1964,, p. 20) writes:

It 1s doubtful whether decisions are generally directed towards a goal.

It is easier and clearer t o view decisions as being concerned with discovering courses of action t h a t satisfy a whole s e t of constraints. It is t h s s e t , and not any one of its m e m b e r s , t h a t is most accurately viewed a s the goal of the action. If we select any of the constraints f o r

(18)

special attention, it is ( a ) because of its relatlon t o the motivations of the decisionmaker, or ( b ) because of its relatlon to the search process that is generating or designing particular courses of action. Those constraints that motivate the decisionmaker and those that guide his search f o r actions a r e sometimes regarded as more "goal-llke" than those t h a t limit t h e actions he may consider, or those t h a t a r e used t o t e s t whether a potential course of action he has designed is satisfactory. Whether we t r e a t all t h e constraints symmetrically or refer t o some asymmetrically, a s goals, is largely a m a t t e r of linguistic or ana- lytic convenience.

When a distinction between objectives and constraints is made, it is usually based on the idea of a constraint as an absolute restriction, in contrast t o an objective or goal that may be open ended. Majone (1978) suggests, and we con- c u r , t h a t when there a r e several objectives they c a n always be traded off a t the margin if this leads t o an improvement in the total utility. That is, it 1s reasonable to sacrifice a particular objective if the situation on the whole is thereby improved. A constraint on the other hand, cannot be so exchanged against other constraints, for its logical force resides wholly in its inviolability. Thus, the translations of a decislonmaker's desires into a problem formulation, includ- ing t h e definitions of limit values (constraints), must be done with a lot of c a r e . For once a constraint, a limit value, is established by, or in concurrence with, the decisionmaker, it ~ 1 1 1 be held to during analysis (although, of course, it may be changed in a later analysis). Majone (1978) r e m a r k s :

The opportunity cost of a proposed policy constraint must be carefully considered before the constraint becomes firmly embedded in the ana- lytic structure. As Hitch and h<cKean ( i 9 6 0 , p . 196) write, 'casually selected or arbitrary con.straints can e a s ~ l y increase system cost or degrade system performance manifold, and lead t o solutions t h a t would be unacceptable to the person who s e t the constraints in the first place.' They cite the example of a weapon-systems study, where a

(19)

constraint on acceptable casualities led to solutions in w h c h 100 mil- lion dollars was being spent, a t the margin, to save a single life. Many more lives could have been saved with the same resources. Had t h e policymakers realized the opportunity cost of their safety requirements, they would probably have s e t them a t a lower level. Or, like good utilitarians, they may have chosen to t r e a t the risk factor as a subgoal, to be optimized compatibly with o t h e r system's requirements and the available resources.

The constraints may be so restrictive t h a t no alternative will attain t h e objective. Demonstrating t h a t s o m e t h n g cannot be done or can only be done if certain constraints a r e removed may be just as important a s showing t h a t something can be done, for it can save a g r e a t deal of wasted effort and be on a sounder basis.

Here again the blood availability example illustrates the points. We have so far treated the desirability of always having blood available when and where needed a s an objective, because the existing system allowed modest shortfalls affecting elective procedures. However, if i t had been considered and treated a s an inviolable constraint-one t h a t could easily be advocated by someone not familiar with t h e usual working of the health system-the analysis and its results would have been violently transformed, as also would its results.

On the other hand, a constraint may be intrinsically inviolable. For example, the IIASA energy study could not yield a use of fossil fuels over the next fifty years g r e a t e r t h a n t h e available supply. However, careful scrutiny of t h e technical possibilities yielded new views of this constraint, a s new forms of fossil resources could be exploited a s fuels when technology and economics become favorable (example: oil from oil sh.ales) o r t h e available supplies could be extended by combining t h e m with m o r e advanced technologies (example: using breeder reactors to fuel coal gasification).

The lesson is simple: objectives and constraints must not be treated as sacredly inviolable, but m u s t be scrutinized from many points of view as t h e

(20)

analysis proceeds and technical possibilities emerge-and thew roles should bc subject to change.

3. ALTERNATIVES

The s e a r c h for alternatives is the activity t h a t gives form and structure t o systems analysis. No amount of modeling and evaluation will help to move t h e effort toward a solution unless the analyst c a n discover, design, or invent one or more satisfactory alternatives, t h a t is t o say, actions or policies that offer hope of accomplishing what t h e decisionmaker wants. No amount of evaluation will uncover the best alternative available unless it appears among those investigated.

An obvious illustration is suggested by Sugden and Williams (1978, p. 231):

Consider, for example, a cost-benefit analysis t h a t compares t h e effects of undertaking a large programme of road-building in a city with the effects of taking no action and w h c h finds t h a t the former policy ls t o be preferred. This might then be used as a n argument t h a t the whole programme ought t o be undertaken even though all that has been shown is t h a t this is better t h a n doing n o t h n g a t all. It might well be that undertaking only a p a r t of the programme is preferable t o either of the alternatives t h a t has been studied.

To generate alternat-ives a t the s t a r t of a study is clearly a creative a c t . Once we have one, it is easy t o design an infinity of related alternatives t h a t a r e more costly or less efficient or possibly even marginally better, but significant improvement is harder t o achieve. One must make a deliberate effort t.o think of possibilities; here brainstorming and talking t o a variety of people helps.

To design alternatives frequently means becoming well acquainted with t h e relevant technologies and working closely with t h e technologists with detailed knowledge of the techrlical possibilities. For example, the flood-control engineers developed the options for protecting the Oosterschelde estuary that were considered by the Rand t e a m of systems analysts. In the IIASA study of

(21)

world energy supply and demand a great deal of effort was expended on detailed inquiries into the possible technical options t h a t might be available for use with many forms of energy sources: coal, oil, solar, ocean t h e r m a l , water power, and so on. The characteristic properties of these technical options could t h e n enter the systems analysis calculations in the forms of contributions t o the total energy supply, costs, material requirements, demands on the available stock of capital, and so on.

T h s is not to say t h a t the systems analyst should just consult a few technologists and then adopt some alternatives. Experience shows t h a t a n interplay between the original alternatives and the issues in t h e analysis may suggest important options with m u c h improved properties in the light of the comprom- ises and tradeoffs t h a t the analysis reveals. For example, although the IIAS.4 study of world energy supply and demand started with t h e presumption t h a t it would be possible for t h e world to move in 50 years from a major dependence on exhaustible sources of energy to principal reliance on renewable sources, the analysis showed t h a t t h s would be so difficult as t o be virtually impossible in so short a period. In f a c t , the study projected a significant increase in t h e use of energy from exhaustible sources (as Table 3.6 shows), although the proportion of the total supply from such sources could be expected t o decrease somewhat (as Figure 3.10 shows). To make m a t t e r s worse, t h e analysts were forced to project a continuing high use of gas and easily transportable liquid fuels. With use of oil and gas from fossil sources having t o decline for both supply and cost-of- production reasons, these find.ings turned attention, a s Figure 3.10 shows, to coal liquefaction a n d gasification, since the world has huge coal reserves.

The lIASA analysis t e a m observed (Energy Systems Program Group 1981):

When coal-based synthetic fuels begin to take on such a large role, the technical processes by which they a r e produced also become most significant. . . .

In autothermal coal gasi-fication and liquefaction schemes, both the process h e a t and the required hydrogen a r e produced by burning

(22)

coal, in addition t o t h e a m o u n t of coal needed for t h e chemical c a r b o n c o n t e n t of t h e synfuels. A large a m o u n t of energy is lost in t h e conver- sion p r o c e s s , and t h e resulting gas o r liquid contains a b o u t half of t h e e n e r g y c o n t e n t of t h e original coal.

For t h e a l l o t h e r m a l p r o c e s s t h e p r o c e s s h e a t a n d t h e required hydrogen a r e supplied exogenously, preferably by m e a n s of h e a t f r o m a n u c l e a r r e a c t o r s u c h a s t h e high t e m p e r a t u r e r e a c t o r ^.^.^.o r in m o r e futuristic s c h e m e s by m e a n s of hydrogen gas from a solar p l a n t . The synfuels t h u s p r o d u c e d have a higher e n e r g y c o n t e n t t h a n t.he original coal. While in b o t h p r o c e s s e s t h e combustion of coal r e l e a s e s c a r b o n dioxide i n t o t h e a t m o s p h e r e , t h e allothermal p r o c e s s r e q u i r e s less coal (by a f a c t o r of 3 to 4) a n d accordingly r e l e a s e s a smaller a m o u n t of c a r b o n dioxide t h a n t h e a u t o t h e r m a l m e t h o d . ^.. ^.

Thus, in t b s c a s e t h e analysis showed t h a t t h e coal gasification a n d liquefaction p r o c e s s t h a t m a d e i m p o r t a n t use of o t h e r energy s o u r c e s , s u c h a s a h i g h - t e m p e r a t u r e n u c l e a r r e a c t o r or hydrogen gas from a solar p l a n t , offers

~ m p o r t a n t advantages: extending the useful life of t h e coal r e s e r v e s , allowing t h e shift from readily t r a n s p o r t a b l e gas a n d liquid fuels t o t a k e place over a longer period, and r e d u c i n g t h e a m o u n t of c a r b o n dioxide r e l e a s e d i n t o t h e a t m o s p h e r e . The detailed calculations showed thls hybrid option. t o have i m p o r - t a n t advantages when development and investment allow lt t o b e i n t r o d u c e d .

There a r e m a n y f o r c e s t h a t t e n d t o r e s t r i c t t h e r a r g e of alternatives likely t o be examined. Some of t h e s t r o n g e s t a r e biases of various s o r t s d u e t o t h e unconscious a d h e r e n c e t o a n organization's " p a r t y line" or c h e r i s h e d beliefs o r even m e r e loyalty (Kahn a n d Kann 195'7). When a problem is first discovered in a n organization t h e r e is a t e n d e n c y t o look for a solution t h a t c a n be controlled w i t b n t h e organization. An a d m i n i s t r a t o r m a y initially b a r a n a l y s t s f r o m considering c e r t a i n kinds of alternatives for no b e t t e r r e a s o n t h a n t h a t "we d o n ' t d o things t h a t way." Staff analysts a r e particularly vulnerable t o biases of t h i s s o r t .

(23)

Another thing t h a t can happen is t h a t the analyst in talking with the decisionmaker or h s staff becomes aware that the decisionmaker or h s superior doesn't like certain kinds of alternatlves. He may get the feeling t h a t it is not only useless but also hazardous to even give the impression t h a t he might be advocating these alternatives a s possible solutions. As a result, t h e development of these alternatives is likely to be neglected or forgotten.

Various alternatives can also be advocated by enthusiasts, e a c h of whom quite honestly believes that this alternative is the royal road to the problem's solution. One could easily imagine, in the blood-supply example, one enthusiast advocated all blood being delivered to t h e hospital blood banks on a prescheduled basis, with another advocating with equal enthusiasm that all deliveries be m a d e on demand. With t h e analysis in hand, we now know t h a t the best compromise is a set schedule supplemented by a few demand deliveries when blood-bank supplies run short-in short, a hybrid option with an average of 4.2 deliveries p e r hospital per week, two-thrds prescheduled and the other one- third on demand. Similarly, any citizen with even passing interest in the energy problem recalls t h a t various enthusiasts advocate singular solutions. One of the most vocal groups feels t h a t solar energy is an early prospect t h a t the world must adopt. On t h e other hand, perhaps the most important finding of t h e IIASA energy systems analysis is that-at least for t h e next 50 years-no one energy option can be relied on t o solve the world's energy supply problem; r a t h e r , a large spectrum of such options, as Figure 3.10 suggests, must be combined appropri.ately, with t h e totality managed with a view t.o having a completely sus- tainable system within the next century. In sum, t h e enthusiastically sponsored unitary solar option cannot do t h e job within the next 50 years, according to the JlASA study.

Alternatives need not represent t h e same approach t o obtaining t h e objective. Suppose, for instance, the objective were to reduce crime. There a r e at least two categories of alternatives that might help t o attain this goal-social measures such as preventive education and antipoverty legislation or- police

(24)

m e a s u r e s s u c h as s e v e r e r p u m s h m e n t and m o r e c e r t a l n apprehension of law n o - lators I n t h e ~nvestigatlon, p o l ~ c e m e a s u r e s a s whole might be one a l t e r n a t ~ v e , o r a s p e c ~ f l c p o l ~ c e m e a s u r e such a s a 50 percent increase In t h e number of pol- Ice c a r s rnlght be a n o t h e r , o r a combination of several police m e a s u r e s with soclal m e a s u r e s mlght be a f u r t h e r alternative

The point is t h a t a lot of thought has to be given t o t h e design of alternatives. They will be eliminated o r modified t o remove o r add c e r t a i n features a s constraints a r e discovered a n d applied, or a s their effectiveness is e s t i m a t e d . Prelinunary evaluation in addition c a n be used to eliminate t h e grossly inferior and those t h a t a r e d o m i n a t e d by o t h e r alternatives. During t h e evaluation process t h e good f e a t u r e s of t h e b e t t e r alternatives m a y suggest ways for t h e analyst to design new and still b e t t e r alternatives.

Once we have a model o r s e t of models t h a t enables t h e analyst t o d e t e r -

mine t h e significant c o n s e q u e n c e s of a class of alternatives, t h e analyst c a n investigate how t h e s p e c t r u m of consequences changes with changes In t h e alternatives (belng careful of course not t o m a k e t h e changes so radical t h a t t h e c r e d ~ b i l l t y of the model is brought into question) a n d t h u s improve some of t h e alternatives The analyst m u s t , however, not t r y t o c a r r y t h s process too f a r , for, although t h e model r e s u l t s may show differences, they m a y really b e inslgm- ficant considering how c r u d e o u r models a r e likely to be. There a r e , of c o u r s e , c e r t a i n relatively narrow p r o b l e m s t h a t p e r m l t a closed m a t h e m a t i c a l formulation o r model where a n algorithm of one s o r t o r a n o t h e r , linear prctzrammlng f o r instance, p e r m i t s t h e analyst t o investigate all alternatives of a certain class and d e s ~ g n a t e one a s best There a r e few, ^lfa n y , sociotechnical problems where such a formulation is possible, however

A s noted in Chapter 4, t h e r e a r e usually a n u m b e r of characteristics o r unstated objectires t h a t a decisionmaker expects o r would like t h e alternatives he selects t o attain in reasonably high degree: insensitivity, reliability, inv.ulner- a b i l ~ t y , znd flexibility a r e s o m e of t h e s e desired char.acter*istlcs.

(25)

The deliberate generation of a wide range of alternatives is an essential step in systems analysis. One cannot investigate all of them thoroughly; it would be too costly and excessively time-consuming. Moreover, one could not then expect the client to look over the results. Some f o r m of screening that reduces the number of alternatives to something like t e n or less should be applied. Dom- inance or almost dominance may get rid of a few. Eliminating those with some important negative impact such as cost excessively large is another possibility.

Elimination based on the analyst's judgment of a group of impacts is a further possibility; one can set up a s e t of standards based on five or six characteristics and those t h a t fail to m e e t these standards c a n be judged not good enough (Eilon 1972).

The final few alternatives that survive t h s screening may then have to be examined in even more detail. In fact, even after the selection is made, there may be aspects of the chosen alternative t h a t need to be worked out in greater detail before the final program for implementation is prepared.

Strong alternatives make for easy analysis. A n outstandingly good alternative may make further analysis unnecessary; it may simply look so superior t h a t acceptance is immediate.

In sum, i t is important to begin the analysis with good alternatives, and t h e time and effort needed to develop them. must be adequate to the task, and all of the relevant sources of technical expertise must be tapped in the process. On the one hand, the analysts should beware of the glib technologist who alleges that there are only two or t h r e e possibilities; on the other, he should be equally suspicious of alternatives developed entirely within the systems analysis t e a m , for it will usually not contain representatives of all of the useful technologies.

Rather, i t is a synthesis that is called for, led by the systems analysis t e a m with its eye clearly on the problem i n hand.

However, most important of all is the continuing awareness of the need for newer and better alternatives throughout t h e analysis that frequently holds the key to outstandingly useful results. The IJASA energy analysis team concluded

(26)

that "The demand for liquid fuels is a principal driving force of t h e energy prob-

;L

lem." Their hybrid coal gasification and liquification scheme is an imporint step toward solving this driving problem.

REFERENCES

Arrow, K . J . 1951. S o c i a l Choice a n d / n d i v i d u a l V a l u e s . New York: Wiley.

Dror, Y . 1975. Some Features of a Meta-Model for Policy Studies. P o l i c y S t u d i e s J o u r n a l , 3, 3 p p 247-255.

Eilon, S. 1972. Goals and Constraints in Decisionmaking. @erat.lonal R e s e a r c h Quarterly, 23,(1) 3-15.

Energy systems Program Group of IIASA, Wolf Haefele, Program Director, 1981 E n e r g y in a F i n i t e World: V o l u m e 1 . P a t h s t o a S u s t a i n a b l e F'uture. Cam- bridge, Xassachusetts: Ballinger Publishing Company.

Hatry, H.P. 1970. Measuring the Effectiveness of Nondefense Public Programs,

@erati.ons R e s e a r c h , 18,5,p 774.

Hitch, C.H. and R . McKean, 1960. T h e E c o n o m i c s of D e f e n s e in t h e N u c l e a r Age Cambridge, Mass: Harvard University P r e s s .

Hovey, H.A. 1968. T h e P l a n n i n g - P r o g r a m m i n g A p p r o a c h t o G o v e r n m e n t D e c i s i o n m a k i n g New Y 0rk:Praeger.

Kahn, H. and I. Mann, i357. T e n C ~ m m o n P L t f a l l s , RM-1937, Santa Monica, Cali- fornia: The Rand Corporation.

Keeney, R.L. and H. Raiffa, 1976. D e c i s i o n s with M u l t i p l e O b j e c t i v e s : P r e f e r e n c e a n d Tfalue T r a d e o f f s . New York:M-iley.

(27)

Majone, G . , 1978. The ABC's of Constraint Analysis, Working Paper No 2, New York:Russell Sage Foundation.

Nagel, S.S. and M. Keef, i979. Policy Ana1ysis:In Social Science Research, Bev- erly Hills, California: Sage Publications.

Raiffa, Howard 1964. Decision Analysis: Introductory Lectures on Choices u n d e r Uncertainty. Reading. Massachusetts: Addision-Wesley.

Simon, H., 1964. On t h e Concept of Organization Goal, Administrative Science Quarterly, Vol 9, No 1.

Sugden R . a n d A. Williams, 1978. The Principles of Practical Cost- Benefit Analysis Oxford, England: Oxford University Press.

Walker, W.E., J.M. Chaiken, and E.J. Ignall, 1979. fire Department Deployment Analysis: A Public Policy Analysis Case S t u d y , New York, Worth Holland.

Wohlstetter, A. 1964. The Analysis and Design of Conflict Systems. In E.S. Quade ( e d . ) , Analysis for M i l i t a q Decisions. Amsterdam: North-Holland, p. 123.