J 3tems Analysis and Operations Research - USAID

Loading...

3tems Analysis and Operations Research: -

J

'A TOOL FOR POLICY AND PROGRAM PLANNING FOR DEVELOPING COUNTRIES

Board on Science and Technology for International Development Commission on International Relations

r

D

1

C

(:oogIe

SYSTEMS ANALYSIS AND OPERATIONS RESEARCH: A Tool for Policy and Program Planning for Developing Countries Report of an Ad Hoc Panel of the ·Board on Science and Technology for International Development Commission on International Relations 11<>.+.0".1 /l1t'''."J,..

(o".. c./.

Con resumen en espaiiol Avec resume en fran~s

NAS-NAE NATIONAL ACADEMY OF SCIENCES Washington, D.C. 1976

APR 1 3 1976

LIBRARY

rt,

I"/G-OJ/(/

(.1

This publication has been prepared by the Panel on Strengthening the Capabilities of LessDeveloped Countries in Systems Analysis of the Board on Science and Technology for International Development, Commission on International Relations, National Academy of Sciences-National Research Council, for the Office of Science and Technology, Bureau for Technical Assistance, Agency for International Development, Washington, D.C., under Contract AID/csd-2584, Task Order No.1. NOTICE: The project that is the subject of this report was approved by the Governing Board of the National Research Council, whose members are drawn from the Councils of the National Academy of Sciences, the National Academy of Engineering, and the Institute of Medicine. The members of the Committee responsible for the report were chosen for their special competences and with regard for appropriate balance. This report has been reviewed by a group other than the authors according to procedures approved by a Report Review Committee consisting of members of the National Academy of Sciences, the National Academy of Engineering, and the Institute of Medicine.

Order from ~ Nationa\ Technical \nformation ~ Springt"ieMt, Va.

=

1ID.tf> '}5..t b3g

PANEL ON STRENGTHENING THE CAPABILITIES OF LE~EVELOPEDCOUNTRIES IN SYSTEMS ANALYSIS

Members

Massachusetts Institute of Technology, Cambridge, Massachusetts, Chairman IRMA ADELMAN, University of Maryland, College Park, Maryland BERNARD RA YMOND BELL, International Bank for Reconstruction and Development, Washington, D.C. AUGUSTINE O. ESOGBUE, Georgia Institute of Technology, Atlanta, Georgia DAVID BENDEL HERTZ, McKinsey & Company, Inc., New York, New York WILLIAM A. W. KREBS, Arthur D. Little, Inc., Cambridge, Massachusetts PETER PHILIP ROGERS, Harvard University, Cambridge, Massachusetts D. V. N. SARMA, Operations Research Group, "Nandanvan," Baroda, Gujarat, India PHILIP McCORD MORSE,

Resource Person ARTHUR A. BROWN,

Arthur D. Little, Inc., Cambridge, Massachusetts

Observer LEHMAN FLETCHER,

Agency for International Development, Washington,

D.C. Staff

Board on Science and Technology for International Development, Commission on International Relations, National Research Council, Professional Associate

B. K. WESLEY COPELAND,

***

Board on Science and Technology for International Development, Commission on International Relations, National Research Council, Head, Special Studies.

JULIEN ENGEL,

iii

Preface

The application of the scientific method to problems of management has received great impetus since World War II. Today, practitioners of Systems Analysis and Operations Research (SA/OR) are to be found at the service of, or working within, government, industry, transportation, health, and many other fields characterized by complex activity. As a formalized practice, SA/OR has spread widely around the world, particularly in the industrialized countries. Does SA/OR have relevance for the problems that typically confront the developing countries? And, if so, how should these countries go about acquiring or strengthening an indigenous SA/OR capability? These questions, among others, prompted the Agency for International Development (AID) to request a report on the subject from the Board on Science and .Technology for International Development (BOSTID). As in the case of other studies undertaken by the Board, it was understood the report would have a dual purpose: (a) to provide guidance to AID and other agencies similarly concerned with providing technical assistance to developing countries; and (b) to inform a broader, nontechnical readership, especially in the developing world, of the essential character, capabilities, and functions of SA/OR. This report, accordingly, attempts to provide the following: • An exposition of systems analysis methodology and its limitations, with emphasis on the research aspects and the practical nature of the results sought;

v

PREFACE

vi

• Illustrative applications of SA/OR in public- and private-sector decision making, with particular emphasis on the problems and needs of developing countries; • Requirements for establishing and operating an SA/OR group and the variety of organizational modes that might be suited to the conditions of the developing countries; • Training processes and criteria for producing competent systems analysis personnel in developing countries; and • Technical assistance implications and opportunities for consideration by AID (and other development agencies). The Board convened a special panel for this task composed of teachers, practitioners, and "consumers" of SA/OR. Some members are from developing countries; others have a substantial background of profeSSional work in the developing countries or intimate association with the product of SA/OR in the formulation of development projects and policies. Early in its deliberations, the panel concluded that its report should be aimed especially at senior officials and executives in developing countries. Since it is the upper-echelon decision makers who are in the best position ultimately to appreciate the value of SA/OR, for their benefit the summary of the report is somewhat more detailed than usual. As chairman of this panel, I take pleasure in saying that the panel members distinguished themselves by their cooperation and dedication to the task. Their contribution reflects the best tradition of the SA/OR discipline. Special thanks are due to Dr. Francisco Sagasti (Lima, Peru) who was unable to accept the Academy's invitation to become a member of the panel, but who reviewed an early draft of the report and, through his several communications to the panel, made a valuable contribution to its deliberations. The panel had the exceptional good fortune of having as its principal resource person Dr. Arthur Brown, an SA/OR practitioner whose experience stretches to the dawn of the discipline. He patiently and lucidly prepared all drafts of the panel's report, endured with good grace the many additions, deletions, and revisions requested by his colleagues, and gave unstintingly of his time in polishing the fmal version. PHILIP MORSE

Chairman

Contents

SUMMARY 2 SA/OR IN THEORY AND PRACTICE DefInitions and Generalities lliustrative Examples Baroda Bus Study: Simulation Approach for Routing and Scheduling World Bank Study: Water and Power in the Indus Basin South Korean National Economic Models Systems Study of Agriculture in Mexico Blood Bank Inventory Control Brief Examples 3 THE APPLICATION OF SA/OR

8 8 10 11 13 17 23 28 32 38

Overview The Process in Detail

38 39

4 ORGANIZING FOR SA/OR

46

Key Personnel Organizational Forms Location of the Group Equipment and Facilities An Initial Work Program Some Caveats

46 47 49 51 51 53 vii

viii

CONTENTS

5 TRAINING SA/OR WORKERS Training the Administrator Training the Director Training the SA/OR Staff 6 TECHNICAL ASSISTANCE IMPLICATIONS Developing University Training Programs Starting an SA/OR Program

57 57 58 59 61 61 63

GLOSSARY

65

APPENDIX A Excerpt from Foreword to World Bank Report on the Indus Project APPENDIX B Bibliography for Table 1 APPENDIX C Supplemental Reading List

71 73 78

RESUMEN EN ESPANOL

80

RESUME EN FRANc;AIS

88

. .

1 Summary

A central problem facing the governments and people of the lessdeveloped countries (LDCs) is the management of profound and rapid change. There is an urgent desire to compress into the span of one generation the process of industrialization that the advanced countries took nearly 10 generations to complete. Slow change, based on trial and error, can and does occur without guidance, as it has in much of the industrialized world-but it is wasteful. On the other hand, when change is rapid, guidance from a number of sources may be required. The greater the change, the greater the need for understanding the factors that promote or oppose it, help or divert it. The changes that accompany technological and economic progress touch every aspect of life and culture: the exploitation of raw materials and energy resources, the training and organization of labor, the role of women, the character of family life, the school experience, the use of leisure time; the list is endless. In an atmosphere of change and uncertainty, planners and government officials need to use all reliable methods for planning, analysis, and control. This report deals with one proven method of helping the administrators in public or private organizations make their decisions. Over the past 35 years there has been far-reaching development in the techniques for applying the ways of thinking and working commonly used by scientists to the problems confronted by decision makers in government, business, and other institutions. These techniques are called, variously, Operations Research, Systems Analysis, Management Science, and Cybernetics. All these terms mean much the same thing and therefore are combined here under

2

SYSTEMS ANALYSIS AND OPERATIONS RESEARCH

the designation Systems Analysis/Operations Research (SA/OR). The notation is clumsy, but there seems to be no better alternative. * In the course of its growth, SA/OR has suffered at times from the exaggerated claims made by some of its proponents. A formal refutation of such claims is difficult, and the panel has not attempted it, since its members believe that a realistic understanding of SA/OR methods and of the areas of application serves to put these claims in proper perspective and is more useful for the report's intended audience. Briefly, it may be said that SA/OR helps the decision maker by working on those areas of his problems that can yield useful quantitative measures and related theories. A basic tenet of SA/OR is that the work must be carried out in close cooperation with individuals familiar with the peculiarities of local customs, and that the final decision must be made by an authority knowledgeable about those peculiarities, not some outsider. The accomplishments of SA/OR, however, are real, and its growth has been significant. The International Federation of Operations Research Societies (IFORS) has member societies in 25 developed and developing countries. The Institute of Management Sciences has individual members in 77 countries. An International Institute for Applied Systems Analysis in Austria has recently been created by agreement among several governments. An extensive body of theory has been built up with respect to many concrete problems and is ready for application by those agencies and enterprises that have not yet made use of it. Current problems of inventory control, production scheduling, fleet management, hospital operation, deployment of community service centers (fire, police, churches, libraries, banks, etc.), urban layout, and many others have been thoroughly studied. Longer-range problems, on a larger scale, have also been attacked, and progress has been made in dealing with the allocation of national resources (as will be seen from the examples cited later in this report). The SA/OR workers are technicians. They do not determine policy; that is the task of the elected or appointed officials-the administrators-who are conversant with local peculiarities, physical and social, and to whom the SA/OR workers report. The workers' job is to point out to the administrator the consequences of various alternative policies and to work out, in concert with him, ways whereby a chosen policy can be implemented. Naturally, the more knowledgeable the workers are about the unique physical and social conditions of the country involved, the more useful and practical will be their reports. For this reason it is important that as large a proportion as possible of the initial SA/OR team be nationals of the country involved. This report indicates a number of ways whereby these nationals can be recruited, trained, and put to work for the benefit of their own countries. *Technical terms, designated by an asterisk, are explained in the Glossary.

SUMMARY

3

SCARCE RESOURCES AND SA/OR SA/OR is particularly useful when scarce resources must be used ef· fectively. Resource-allocation problems are worldwide. They affect the devel· oped as well as the developing countries. Historically, however, the developing countries were largely confmed to the production of one crop or raw material, or, at best, a few. Today they want to make general use of all their resources. They find that at any given time some factor (such as skill, manpower, energy, transport, material, or capital) is in critically short supply and that the scarcity impedes progress. Under these conditions, LDC governments and planners are constantly faced with the need to make best use of the resources currently available, to plan for the conservation of resources that will be needed in further stages of progress, and to balance current needs against investment for the future. This is the domain of Systems Analysis and Operations Research. SA/OR contributes to current efficiency and improvement of the planning process in three ways: • Through the development of models· and analogues, it provides an understanding of the relevant factors and of the ways they interact. • By calculation and by simulation,· it predicts the possible consequences of alternative actions and policies. • By calculation, planned experimentation, and simulation, it assists in the selection of the best path to the desired goal.

In the past two decades there have been exaggerated claims of the potential scope of SA/OR. It is essential to note that the value of SA/OR lies in its facilitation of the planning process and its assistance in the control of operations and development. It is not a substitute for political judgment, but assists by defming the boundaries of the possible and by predicting the results of actions, or stating the limits of certainty for such predictions. There are, of course, cases in which the use of SA/OR is inappropriate; these will be discussed in the section on "Caveats." Five examples of applications of SA/OR are discussed in later sections of this report: the development of the Republic of Korea's Second Five-Year Plan; the World Bank Study of the Water and Power Resources of West Pakistan;t a systems study of crop agriculture in Mexico; a study of bus scheduling in Baroda, India; and a U.S. study of efficient management of whole-blood inventories for hospitals and groups of hospitals. Other examples can be found in the planning of agricultural development, design of seaports, man· agement of transport systems, location of distribution facilities (warehouses,

t Since 1971 West Pakistan has been known as Pakistan.

4

SYSTEMS ANALYSIS AND OPERATIONS RESEARCH

collection stations), scheduling of production, and in many other areas. These are briefly noted in Table 1, p. 34. (See also Bibliography, p. 73.)

PRACTICING SA/OR IN LOCs The requirements for the successful practice of SA/OR are not excessive. A working SA/OR group can consist of as few as four to six professionals, as long as one is skilled in the use of electronic computers and another in applied mathematics; the remaining members may be scientists, economists, or mathematicians. The group should be headed by a director and should report to a decision maker (in this report also called the administrator) who is not a part of the group but who uses its findings and collaborates in its work. The effectiveness of the group will be determined by the willingness of the decision maker to discuss his problems freely and to give counsel. Success also depends largely on the competence of the group's director, who must share the concerns of the administrator and must understand the methods of his co-workers. He must also have the imagination to ask new questions and the ability to apply the ideas and talents of his group to ensure that they will be put to maximum use. An SA/OR group will need to use consultants to supplement its technical abilities-for instance, civil engineers when dealing with seaport design or road networks; econometricians when dealing with national plans; and experts in fields such as agriculture, banking, ecology, and public health when dealing with regional development. Such consultants may be recruited locally or abroad, as needed for specific projects.

The group must have access, whenever necessary, to a large, fast computer. (The computer can be located anywhere, even in a foreign country.) If such access is assured, the group will not need to own a large machine. Much experimental work and much data preparation can be done on small machines of modest capital cost.

TRAINING FOR SA/OR A knowledge ,of statistics and probability theory, advanced calculus, linear algebra, economics, and some branch of natural science is required in most U.S. undergraduate programs for SA/OR students. Computer programming and social science courses are also desirable. Graduate programs usually require mathematics, allocation theory,· applied stochastic processes,· management of research and development, portfolio and financial management, site selection,· long-range forecasting, queuing· and inventory theory, trans-

SUMMARY

5

port, traffic and parking systems, water resource management, health care and social welfare delivery systems, and educational and urban systems. While some LDC universities under European influence have traditionally stood apart from practical work, there is increasing recognition that faculty and students need to be challenged by real-life problems. The development of SA/OR capabilities can help promote this trend in the following ways: • Through local and regional conferences, university faculties can be exposed to the professional opportunities in SA/OR and to the needs of their countries. • SA/OR groups can seek out faculty members for consultation and for the development of training courses in which practical problems are made part of the curriculum. • A fellowship program for study or work with SA/OR groups in industrialized countries can create an awareness of the practical role of the scientifically oriented faculty member. Both the prestige and the competition associated with the fellowship program can help to promote an awareness of the nature and value of SA/OR.

OBSTACLES TO DEVELOPMENT AND USE OF SA/OR Perhaps the principal obstacle to the use of SA/OR is the administrator's lack of familiarity with its nature and applications, which is typically reflected in statements ranging from lukewarm to negative. Sagasti reports statements such as the following from government officials: 1 • Government institutions are complex and the bureaucracy is very slow. • There are many political problems involved; they do not lend themselves to solution by the use of SA/OR. • SA/OR could be useful, but not now. It is too costly, and the data needed for its application are not available. He found the following statements representative of the attitudes of industrial executives: • The only urgent problems are purely technical. • The risk of failure is high, and a guarantee of successful results is needed. • It is obviously impossible for SA/OR to succeed without specific

1 Francisco R. Sagasti, Ma1lllgement Sciences in an Underdeveloped Country: The Case of Operations Research in Peru. (Philadelphia: Management Science Center, Unlversity of Pennsylvania, 1971)

6

SYSTEMS ANALYSIS AND OPERATIONS RESEARCH

expertise directly related to the application. There is a dan&er that the initial applications will be poorly chosen, and that the results will therefore be discouraging. Part of the difficulty is that the entire SA/OR process is often confused with the specific techniques used in the process. The potential user often assumes that SA/OR means linear programming· (or simulation, or queuing theory, or whatever), and he knows his problem is not amenable to attack by that particular technique. Conservative attitudes and resistance arising from unfamiliarity can be overcome only by visible, credible, local examples of successful work. In most LDCs, therefore, SA/OR is likely to grow slowly at first and develop rapidly over a period of 10-20 years. This is in fact what happened in the industrialized countries. Another obstacle may be the shortage of suitably trained practitioners. This is less serious; SA/OR workers can be trained in a fairly short time. Foreign specialists may be retained for the early stages of program implementation while the local group is being trained to assume full responsibility. When choosing such specialists, due care must be taken to ensure that they have had actual experience in SA/OR and that they are able to transfer their experiences by guiding trainees. Once the concept of SA/OR is properly understood, it becomes apparent that lack of data is not an insurmountable obstacle. The first formulation of the research work will define the data needs and show what kind of collection effort is required. Where data are truly lacking, even the crudest of observations will often yield valuable results. In other words, a small amount of data is better than no data.

POTENTIAL BENEFITS It has often been argued, or feared, that resources of skill and money devoted to SA/OR will compete with other essential uses for the same talents and funds and that the opportunity cost of SA/OR will be high. If the use of SA/OR is looked upon as merely the conduct of isolated studies, in which administrators and executives do not participate and which result in reports to decorate library shelves, the gloomy view is justifiable. If, however, SA/OR is undertaken as a cooperative effort of scientifically oriented research workers and action-oriented officials, the saving of resources will more than justify the investment. The application of SA/OR techniques will result in better uses of engineering skills, development funds, capital investments, and human efforts.

SUMMARY

7

FINDINGS The panel's considerations of the applications of SA/OR in LDCs lead it to the following conclusions: • SA/OR is a demonstrated technique for assisting administrators and other decision makers in solving suitable problems. • It is of particular value when limited resources must be used to maximum effect. • Its application is within reach of most developing countries. • It can be developed and implemented by local scientists and technicians. • The industrialized countries can assist LDCs with the training of SA/OR personnel and the development of the SA/OR operation. • SA/OR is not a panacea for the problems of the LDCs; its limitations must be taken into account (see "Some Caveats," p. 53).

RECOMMENDATIONS The panel recommends the following for the attention of developing countries: • Set up a prototype SA/OR group that will report to an administrator committed to its success. • Exercise caution in selecting specialists as advisers, using competitive sources and the advice of experienced SA/OR practitioners. • Encourage university courses in the practice of SA/OR and related subjects. • Sponsor conferences to familiarize business and government executives with SA/OR. • Sponsor scholarships, at home or abroad or both, for potential SA/OR practitioners. The panel recommends the following for the attention of agencies providing assistance: • Encourage the formulation of integrated project plans based on SA/OR study. • Provide funding for scholarships and fellowships in SA/OR. • Assist in the promotion of national and regional conferences on the theory and application of SA/OR. • Support local and regional SA/OR societies. • Give direct support to selected SA/OR projects and programs.

2 SA/OR in Theory and Practice

DEFINITIONS AND GENERALITIES It is the panel's opinion that the typical user, whether in a developing country or elsewhere, need not concern himself with the distinctions among systems analysis, operations research, management science, and cybernetics. Although the debate on differences and merits may be interesting to practitioners, the user's interests are not affected by its outcome. For this reason, the abbreviation SA/OR has been adopted to stand for any or all of the four terms. The question is: What are the common elements in the practice of these disciplines?

SA/OR Is Scientific Research Applied to Operating Problems Reduced to its simplest terms, SA/OR means putting people with scientific/quantitative training to work as applied researchers on the operational and resource-allocation problems of organizations and enterprises. Its practitioners are trained in probability theory, higher mathematics, engineering, computer usage, econometrics, economics, and, to some extent, other social sciences. SA/OR does not use standard techniques to fit standard situations, but actively employs a number of research techniques to find solutions to nonstandard problems. In general, the best results are obtained by

8

SA/OR IN THEORY AND PRACTICE

9

ongoing efforts rather than by one-shot studies; in this sense SA/OR is a continuous process.

SA/OR Uses the Systems Approach The systems approach conceives of entities as wholes marked off from the environment by boundaries and containing subparts whose interactions are conditioned by the presence of the other subparts (subsystems). The action of the whole can be understood only by knowledge of the interactions of all the parts. For example, an irrigation system, considered as a physical entity, con· sists of three or four subsystems: storage, transmission, distribution, and perhaps extraction of water from underground aquifers. This system interacts with the agricultural system, of which it is a part, and with other systems such as general transport, recreation, power, and conservation of land and wildlife. The work of SA/OR is to select a system or subsystem to be analyzed; to defme its boundary; to identify the elements, components, and subsystems; and to develop models that describe the interaction of these components. The immediate purpose of SA/OR is to gain an understanding of the system; the ultimate purpose is to gain control. (For instance, the Port Authority of New York found that traffic through a vehicle tunnel is speeded up if some of the vehicles are temporarily prevented from entering so that platoons can be formed. A platoon is a group of cars following a lead car, which is separated from the last car of the platoon ahead of it; in other words, platoons are clumps of cars that are separated by gaps. It has been found that platoons travel faster than long continuous lines of cars.) Once the system is understood, one or another aspect can be brought to its desired extreme (optimization) by the control measures that can be brought rationally into play. (In the tunnel example, the length of the platoon can be related to the number of cars arriving at the tunnel mouth per minute so that the throughput is as great as possible.)

SA/OR Is Characterized by the Scientific Method The decision maker needs intuition, will, judgment, logic, experience, and rationality. The panel's advocacy of SA/OR does not diminish nor deny the necessity for these qualities.

10

SYSTEMS ANALYSIS AND OPERATIONS RESEARCH

SA/OR, however, adds a new dimension to decision making that is commonly termed the "scientific method." This is an overworked term, almost a cliche, but the panel found no convenient substitute. There is a major difficulty in discussing the scientific method, arising from the fact that the usual characterization makes it sound like the mere exercise of common sense and reason. In fact, the distinctive features of the scientific method are its insistence on measurement and on the use of conceptual models described in quantitative terms, its insistence on experimental verification of its theoretical predictions, and its awareness that its concepts are conditional and subject to change and growth. These characteristics are the basis for the usefulness of SA/OR and the sources of its limitations. The scientific method provides a kind of feedback control for the administrator who can compare what is actually happening with what the administrator believes is happening or should be happening. By its insistence on going to the facts at the outset, rather than beginning from a priori notions (which usually arise from incompletely perceived or incompletely examined experience), SA/OR helps management stay in touch with reality. The SA/OR way is quite different from a number of reporting systems, in which facts may be misunderstood, suppressed, distorted, or honestly confused.

ILLUSTRATIVE EXAMPLES The following examples are offered as illustrations of the methods and values of SA/OR. They are illustrations rather than proofs; proof is difficult to obtain because "success has a hundred fathers." SA/OR is only one component in the mixture of talents and methods needed to fmd and carry out a successful program. It should be emphasized that all of the examples included in this report involved the close participation of local experts and administrators familiar with local customs, social constraints, and boundary conditions. The first example is concerned with a simple but important problem: obtaining maximum use from a fleet of passenger buses. The second deals with a wider area: developing the water and power resources of an entire river system. Like the first, it illustrates the uses of simulation and introduces the techniques of input-output analysis and linear programming. The third example has the greatest scope: it deals with the entire economy of a developing country. The fourth comes back down the scale to consider a single sector of an economy-agriculture. The fifth is limited further to a narrow but vital area: the supply of whole human blood for medical purposes.

SA/OR IN THEORY AND PRACTICE

11

Baroda Bus Study: Simulation Approach for Routing and Scheduling Baroda, India is a fast-growing city.2 Half a million people live in the 78-square-kilometer area. The establishment of an oil refmery, a fertilizer plant, and a petrochemical complex on the outskirts of the city promises faster growth in the future. While the urban sprawl continues, bringing into its ambit more and more industrial units, the trading and commercial center remains in the old city area, where over 70 percent of the population live. The GUjarat State Road Transport Corporation operates bus services in Baroda. The fleet consists of 100 buses, and the route length is about 3,000 kilometers. In 1972 these buses carried 138,000 passengers daily, for a load factor of about 47 percent. Over a period of years the corporation has found that its route length is increasing sharply (through extension of services to the newly developed areas), its load factor is decreasing, and the ratio of peak to average traffic is increasing. Deployment of the limited number of buses, especially during peak hours, is a serious problem. In 1972 the corporation addressed this problem to the Operations Research Group, a private Indian consulting firm, requesting a study of the rationalization of existing routes, peak-hour traffic and its clearance, and the development of schedules and frequencies using the existing bus fleet. 3

Definition of Objectives

It was decided that the interests of the corporation would be best served by the development of methods for generating, testing, and evaluating alternative schedules. Further, it was decided that the aim should be to produce satisfactory, rather than optimum schedules, since the research for the optimum is not only time consuming, but in many cases unrewarding.

Development of the Model

The general bus-scheduling problem was divided into the following areas:

Determination of the demand, i.e., of the number of people who want to 2 For a similar study of the bus system in India's capital city, see Jon Tinker, "How Delhi Made the Buses Run On Time," New Scientist (9 Jan. 1975):64-{)6. 3 For a complete account, queries should be addressed to Operations Research Group, "Nandanvan," Race Course Road, Baroda 7, Gujarat, India.

12

SYSTEMS ANALYSIS AND OPERATIONS RESEARCH

travel between each of all the possible pairs of origins and destinations at different times during the day; Generation of bus routes by a rational calculation, using the demand data; Generation offrequencies for each of the proposed routes; Allocation ofindividual buses to the trips required by the routes and frequency schedules; and Assignment of crews to buses. Finding an optimal solution would require treating all these parts simultaneously, but a satisfactory solution can be found by treating each one separately. The Operations Research Group and the corporation agreed to do the latter. The following activities were therefore undertaken. Data Collection Bus stops were classified into three groups: 1. Those generating significant demand throughout the day; 2. Those generating significant demand at peak hours only; and 3. The remaining stops. The selection and classification were done in collaboration with the staff of the corporation. A detailed observation of the passenger demand at each of the bus stops in the first two classifications was conducted for two days. Observations were made and recorded at five-minute intervals to determine the length of the waiting lines; the arrival, destination, and activity of each bus; and the destinations of the passengers. Use of the Data The raw data were summarized to obtain total boarding demand for each bus stop, on an hourly basis, and were converted into ratios yielding destination demands. The outcome was an origin-destination (O-D) matrix, a two-way table showing the hourly demand for transport to each destination from each origin. Generation ofBus Routes The hourly O-D matrix was scanned to identify the O-D pairs generating heavy flows; these were taken as potential terminals. Other pairs were combined to form feasible routes. Determination ofFrequencies The trip frequencies were generated by a computer program, using a small electronic computer. Without going into elaborate detail, the process can be described as follows: The inputs to the system are the routes (the lists of stops), the travel times

SA/OR IN THEORY AND PRACTICE

13

between stops, and the O-D demand data. Given the route, the demand from each stop is accumulated for every five-minute interval. Trips are initiated on the basis of the following criteria: • The load for the trip shall exceed some minimum level that is prescribed in order to obtain an acceptable load factor. • The total waiting time for the passengers shall be less than the time set by the corporation's service standards. A trip is initiated ifthe load is sufficient or, should the load be insufficient, the waiting time will be exceeded unless the trip begins. This method tends to generate too many trips, in the sense that some passengers will be serviced by many overlapping routes. A second computation, using a much larger computer, is employed to eliminate such redundancy. The first computation provides the outline, and it is articulated in the second. Construction ofSchedules Given the list of trips and routes, it becomes necessary to assign buses (by individual serial number) and to determine the number needed to operate the routes. This was done by a computer program that assigned buses in such a way as to minimize idle time and the number of buses running empty between terminals. Simulation As a test, the operation of the system was simulated by the creation of random demands at the stops that imitated the O-D matrix. Conclusion In an abstract sense, the solution is not the best possible. Nevertheless, it is of great value to the corporation staff because it makes reo scheduling relatively easy and produces workable schedules that meet the main requirements. Initial runs with the computer programs have identified substantial gains that can be realized. The load factor is expected to increase by about 11 percent; experience elsewhere has shown that bus utilization can be increased, and the number of buses required can be decreased by about 10 percent. The recommendations of the study have been implemented in Baroda with results as anticipated.

World Bank Study: Water and Power in the Indus Basin The preceding study dealt with a resource that is in short supply because of fmancial rather than physical constraints; although some expense would be incurred, the supply of buses could be increased if necessary. This

14

SYSTEMS ANALYSIS AND OPERATIONS RESEARCH

study deals with more severely limited (and in some cases irreplaceable) resources-water, land, farming inputs, entrepreneurial skills, and capital. It also involves building a physical structure that requires a high initial outlay, and that, once built, has a limited capacity (the thermal and hydroelectric generating capacity, the electricity-transmission system, and the irrigation system).4 The study illustrates several of the normal features of SA/OR. This one is not a predictive study; it is a study that works out the project design of a complex system. The foreword to the study by Pieter Lieftinck is particularly recommended; it is reproduced in part as Appendix A. The close cooperation between the Pakistani authorities and the SA/OR team is especially noteworthy.

Background

The Indus River and its six main tributaries-the Kabul, Jhelum, Chenab, Ravi, Beas, and Sutlej-form one of the greatest river systems of the world. In the Indus Waters Treaty of 1960, the waters of the Indus, Jhelum, and Chenab Rivers were assigned to Pakistan, and those of the Ravi, Beas, and Sutlej were assigned to India. Pakistan was to receive aid from the World Bank to construct canals that would serve the eastern portion of West Pakistan formerly served by the now diverted Ravi, Beas, and Sutlej. The amount initially allocated for assistance was US $895 million. Subsequently, the donor and lending agencies agreed to provide another US $315 million to meet cost overruns; at the same time the program was limited to the basic essentials. In this connection, an agreement was reached on priorities: the first was to construct an earthfill dam at Mangla, on the Jhelum, with link canals, barrages, and other works specified in the 1960 agreements; the second priority was to conduct a study of the water and power resources of West Pakistan. Remaining funds would be used either to meet the foreignexchange costs of a dam on the Indus near Tarbela if the study showed the dam to be justified, or for other special water projects if the study found them to be more worthwhile. The study was concerned with the best use of large but limited resources; its fmdings would be of major consequence.

4 For a full account see P. Lieftinck, A. Sandove, and T. Creyke, Water and Power Resources of West Pakistan-A Study in Sector Planning. World Bank Study Group. 3 vols. (Baltimore: Johns Hopkins Press, 1968)

SA/OR IN THEORY AND PRACTICE

15

Objectives

The general purpose of the study was to provide the Government of Pakistan with a basis for development planning in the context of successive fiveyear plans. The particular concern was to assist in the formulation of a sound program for the systematic exploitation of water and power resources in West Pakistan. The study was to be an initial phase of ongoing research. It would serve to determine which of the several potential water and power projects were economically viable and feasible for execution during the 1965-70 and 1970-75 five-year-plan periods. It would also serve as a useful guide to the possible future development of water and power projects beyond 1975.

Central Problem and Definition of the System

The complexity of developing an integrated program of power, irrigation, and agriculture for West Pakistan arose from many points of overlap and interdependence, not only within each sector but also among the sectors. For instance, the economics of the large, multipurpose Tarbela Dam project were critically dependent both on the dam's irrigation benefits and on the net power benefits that could be derived from it. But to determine the advantage of Tarbela over an alternative program for power without Tarbela, it was necessary to ascertain the scarcity value of thermal fuel and foreign exchange in the future. Furthermore, the large natural gas resources, suitable as an alternative for power generation, had additional uses, such as fertilizer production. The desirability of fertilizer production, in tum, depended on how the choice was made between water and other inputs for increasing agricultural production. In water-resource development four modes were considered: canal enlargement to enable increased diversion of natural flow; interseasonal surface storage; private tubewell pumping; and public tubewell pumping to balance recharge. Though interdependent to some degree, each had to be separately examined. Finally, the total power-load forecast was heavily dependent on programs for agricultural development, especially tubewell pumping and manufacture of agricultural inputs. In view of all the sectors involved, it was necessary for the study to en· visage the total economy of West Pakistan as a system within which the agricultural and irrigation subsystems and the energy-use subsystem were singled out. The environment was viewed as the rest of the world, as consumer of exports, supplier of imports, and source of foreign exchange.

16

SYSTEMS ANALYSIS AND OPERATIONS RESEARCH

Development of the Models The economy of West Pakistan was modeled by a linear intersectoral model, that is, a system of equations in which the input demands, or consumption, of anyone sector are represented by multiples of its output. (This is an instance of the Leontief input-output· matrix technique.) Two models of the power system were developed: one to forecast the load on the hydropower stations and to calculate the relationships between irrigation uses and power uses; the second to select an optimal power program using net present value as the criterion. The first model was based on computer simulation of the operating regimes of the power network (time variation of the generating and transmission loads). The second model used linear programming techniques. The first model was particularly useful in evaluating the overlap and interdependence between the power and irrigation sectors. For example, it quantified the cost and benefit to power and agriculture of releasing more water over the year as a whole, according to a predetermined release pattern, as against retaining more water in the reservoirs throughout the year. The fundamental concept of integration in the use of surface water and groundwater was at the heart of the proposals that came out of the Indus study. To reach this position required developing a series of water-simulation models that would reproduce the use of the aquifer as an underground reservoir, combined with the transfer of water from "wet" to "dry" areas, and from fresh to saline. One model was based on a breakdown of the basin into the agricultural sector and the attainable cropping intensities that would result from alternative water-development patterns. Another model was developed to determine the live storage that would be required of the reservoirs included in the program as a function of the water requirements of each of the areas analyzed. This model used a manual simulation; that is, electronic computers were not required. Next, an extensive simulation model of the operations of the entire basin was used to test the way the proposed development plan would operate in light of the actual historical hydrological records. This model supported the study's conclusions that the program, if operated as proposed, contained adequate facilities to meet the projected agricultural and power demand. A linear programming optimization model of the agricultural sector was produced to test the economic efficiency of the proposed projects. Since each project uses scarce resources, it is necessary to use linear programming, rather than a ranking by rate of return, so that all resources will be used to give the greatest total return. A project with a very high rate of return may use too much of a scarce resource to permit the inclusion of other projects

SA/OR IN THEORY AND PRACTICE

17

with lesser rates of return that also use the scare resource and that could, in total, yield a greater return on the total investment. For the purpose of this linear program the basin area was broken into the same natural territorial subdivisions used in the other computations to select candidate projects. The canal command areas, covering 33 million acres of irrigated land, taken together constitute the largest single irrigation system in the world. For each designated zone, various combinations of waterdevelopment activities were simulated and their relative advantages were analyzed in tenns of estimated irrigation requirements. A series of hypothetical water-development schemes were dermed and tested for maximum contribution to the expected returns from the irrigation system as a whole. About 500 projects were considered for execution in two time periods. These projects were selected on the basis of the simulations as candidates for the optimization program. The optimization program accepted constraints on the availability of foreign exchange, surface water, public-development funds, and implementation capacity. The linear programming technique simultaneously revealed the effects of alternative levels of availability of these constrained inputs, and it generated an optimum, internally consistent water-development program for the whole basin. To date, investment directly related to the work of the Indus Basin Study has amounted to more than $2 billion. All of the canals and most of tile other works have been executed on schedule and have produced positive benefits.

Summary Evaluation

This pioneering study made an important contribution: it made systems analysis and operations research a valuable tool for regional planning. The planning procedure undertaken by the basin-development study group was a significant step forward from the traditional return-on-investment approach in which programs consist of projects developed and evaluated in isolation from each other.

South Korean National Economic Models While the preceding study dealt with a portion of the economy, this one deals with the economy as a whole. It proVides an example of the way in

18

SYSTEMS ANALYSIS AND OPERATIONS RESEARCH

which SA/OR can pervade an entire national economic planning process. 5 Although the description presented here is very sketchy, some understanding of the SA/OR method can be obtained.

Background In 1964 and 1965 the Korean Government undertook a series of bold reforms in monetary, fiscal, and trade policies. The authorities were prepared to take further decisive steps to break out of economic stagnation. The Second Five-Year Plan (1967-1971) was intended to embody these aspirations and the necessary concrete measures. Nevertheless, at the time the second plan was prepared the economic picture was strongly colored by the poor performance of the first postwar decade. One of the first major tasks specified in the plan was to determine the highest feasible growth rate. The major obstacles appeared to be insufficient domestic savings and inadequate export earnings. As the plan outlines emerged, constructed with aggregate and sectoral consistency tests, it appeared that an annual real growth rate somewhat in excess of 8 percent was possible. Recent history dictated a cautious approach, so a target of 7 percent was adopted and emphasis was placed on exceeding that figure if possible. The Korean Government found a willing ally in the Agency for International Development (AID), which was prepared to argue for higher levels of U.S. assistance to Korea if a technically sound and convincing planning document could be formulated. Toward this end, AID provided a resident mission of advisors.

Organization To secure better coordination between program formulation and implementation, the Korean Government had reorganized its planning mechanisms

5 The principal reference is Practical Approaches to Development Planning: Korea's Second Five-Year Plan, edited by Irma Adelman. (Baltimore: Johns Hopkins Press, 1969) See also the following: Princeton Lyman and David C. Cole, Korean Development: The Interplay ofPolitics and Economics. (Cambridge: The Harvard University Press, 1971) Roger D. Norton, "Planning with Facts: The Case of Korea," American Economic Review (May 1970):59-64. Larry E. Westphal, Planning Investments with Economics ofSeale. (Amsterdam: North-Holland Publishing Co., 1971) Drs. Adelman and Cole were among the principal advisors to the Korean Government on the Second Five-Year Plan.

SA/OR IN THEORY AND PRACTICE

19

by unifying the planning and budgetary functions in an Economic Planning Board headed by the deputy prime minister. The analytic planning work was a joint effort by the Economic Planning Board and AID. Support was provided by other ministries, the Bank of Korea, the Korea Development Bank, and Korean academicians. The only modern computational device routinely available to the planners was an accounting machine. Although some use was made of large electronic computers (available then in Japan, now in Korea), much of the analysis was done with desk calculators and abacuses.

Definition of Objectives

The broad aims of the planning process in Korea may be described as: • Articulation of the society's economic goals; • Calculation of a feasible growth path that goes far toward satisfying these goals; • Identification of major constraints and ways to alleviate them; and • Construction of detailed investment programs and public policies to move the economy along the preferred path. Planning was to be a continuous process; explicit provision was made in the Second Five-Year Plan for a series of annual plans that would detail more specific components and ensure review and revision of the overall design. The first aim, setting economic goals, was not a routine exercise. In the immediate postwar period, Korea had experienced what can only be called national despair. Most Koreans felt their nation would forever remain destitute, that there was no hope of ever coming to terms with what appeared to be a western-style world. At that time, simply establishing goals for change that could be backed up by technical analysis was an important step. The planning process itself was aimed at establishing consistent, not optimum, programs and policies. This meant that programs and policies across the economy and at all levels (aggregaje, sectoral, and project) should be mutually consistent with regard to production levels and input flows.

Development Models

The major quantitative tools employed were:

• An input-output consistency model; • A medium-term macroeconomic· model;

20

SYSTEMS ANALYSIS AND OPERATIONS RESEARCH

• A short-tenn stabilization model; • A mixed-integer linear programming model for the steel and petrochemical sectors; and • A linear programming model for regional balancing. The input-output model contained 43 domestic producing sectors and 34 import sectors, plus 4 value-added and 7 fmal-demand columns. The inputoutput coefficients were based on 1965 prices, modified by technological predictions developed by committees of industry experts. The structural coefficients were based on sample survey data combined with detailed infonnation on production costs, fixed assets, capacity, and output; these were compiled by a series of interministerial "industry committees." Each committee was composed of engineers, fmancial analysts, and economists familiar with the finns in that sector. Among their other duties, the committees acted as review boards for the input-output forecasts. For example, the steel-sector estimates of output, broken down into sales by sector, were reviewed by representatives of each consuming sector; the steel industry's input requirements were checked by specialists from each supplying sector. The fmal demand projections (e.g., consumption, investment by sector, competitive imports, and exports) were determined mainly by a macroeconometric model. In most cases, alternative target levels were established based on selected assumptions. The broad targets were then broken down to the commodity level, by committee judgments. Predetennined criteria were used to select from a set of six alternatives identified by the planning process. Each alternative was based on different assumptions about the behavior of the controllable variables (for example, on the basis of different policies regarding import substitution).

Plan Strategies The four key elements of Korea's long-tenn planning strategy were clearly stated in the second plan: export expansion, domestic-capital mobilization, efficient utilization of manpower and technology, and continuing stabilization of the economy. For the medium tenn, the input-output model exercises had confirmed that exports and domestic savings were limits to growth. Korean planners were convinced that the only bottleneck to rapid export expansion lay on the supply side; to fulfill both the export and employment/technology strategies, investment planning became important. Financial planning of a sort was adopted to assure fmancial stability and to make a closer detenni· nation of the savings constraint. Investment allocation and financial inter-

SA/OR IN THEORY AND PRACfICE

21

mediation became the analytic focus of the first three annual overall reo sources budgets under the second plan. At the beginning of the plan-formulation process, very few proposals were made for viable investment projects. A substantial effort was required to develop each worthwhile proposal. Therefore, the input-output model was used to indicate priority sectors for project (proposal) formulation. Sector-by-sector assessments were developed, as illustrated by the following excerpt from the flfSt annual overall resources budget: Paper products. Investments in this sector will be increased considerably in the future. There is an urgent need for new investment in addition to the projects already in progress.... Nonferrous metals. There are sufficient project proposals in smelting and refining of nonferrous metal ores, but more proposals are needed for p'rocessing facilities such as extrusion and rolling plants. These kinds of statements were based on careful examination and revision of input-output model projections and from tabulation of ongoing major investments. To implement sectoral investment projections made by the plan at the 43-sector level and to disaggregate them into projections for about I SO manufacturing sectors required unusual efforts to gather new technical data. A "man-machine" interaction was established: industrial experts contributed to revision of the technical coefficients in the model and the model projections of interindustry flows contributed to revision of the experts' estimates of future demand. For the continuing annual work on investment guidelines, further cooperation was elicited from the industry committees and from the commercial banks. While the sectoral investment programs were designed to establish consistency on the interindustry demand side, flow-of-funds projections were made for the supply side (savings) consistency tests. The working hypothesis was that identity of savings and investment in real, aggregate terms was no guarantee that the investment programs could be financed in a noninflationary manner. Savings take various forms, and each was appropriate only for certain types of investment. To some extent, the government and private sectors competed for claims on fmancial assets. That competition was reflected partly in the distribution of fmancial intermediation chores between the central bank and the commercial banks. Furthermore, two of the Korean planners pointed out that since the capital market was relatively underdeveloped, other techniques of savings intermediation (perhaps apart from flSCal means) were extremely unreliable. Consequently, the banking sector was heavily burdened with the fmancing of major investments using domestic capital. They also said that this situation implied a stronger interdependency between monetary controls and investment programs in lessdeveloped countries than in many industrialized countries.

22

SYSTEMS ANALYSIS AND OPERATIONS RESEARCH

In the second annual overall resources budget, alternative flow-of-funds projections for the year were presented for different assumptions about monetary policies and exogenous variables. The inflationary implications of each program were explained in accompanying documents. The basic tool for these studies was the stabilization model-a set of estimated equations linking the fmancial and other sectors.

Economic Results Of course it is impossible to disentangle the effects of the plan itself from the various other forces that were in operation, such as the momentum of the preplan policy reforms and the cumulative impact of many years of rather high levels of foreign assistance. Nevertheless, the plan and associated analyses and discussions clearly were a central part of the decision-making process while the Korean economy made its spectacular leap to higher rates of growth_ (Regrettably, these achievements were not associated with greater political participation.) Specifically, with the aid of the plan: • Higher levels of foreign assistance were justified; • The emphasis on foreign trade was turned around, from import substitution to export expansion, with a concentration on labor-intensive industries; • An annual plan-budgeting process was established that helped channel more resources to bottleneck sectors, such as electric power and cement; and • The interdependence of trade, investment, and monetary policies became an accepted part of the decision framework of Korean authorities. During the plan and afterward, the Korean economy continued to grow at about 10 percent per year in real terms, and the dollar value of export ~amings increased by 40 percent annually (before the oil price increases of 1973 and 1974). The real income of the poorest segment of the population increased about as fast as total income; underemployment diminished markedly.

Other Uses of SA/OR Models

In the period immediately following publication of the plan, a mixedinteger programming model was used to test the validity of two major investment proposals-to build a petrochemical complex and to build an integrated steel mill. Each would compete for large shares of the resources of the economy. The result was that the steel mill was postponed and the

SA/OR IN THEORY AND PRACTICE

23

petrochemical complex delayed until demand necessitated the building of a larger, and hence more economical, plant.

Conclusions

Perhaps the most pervasive consequence of the work resulting from the second plan is that quantitative tools of economic analysis are now taken for granted in Korea and employed to study a variety of issues. By the end of the second plan period, the numerical forecasts in the plan document had become outmoded because of forecasts that, ironically, had been too conservative. Nevertheless, many of the same kinds of tools of analysis continue to be utilized. That the plan was formulated analytically gave it more credibility in Korea; prior plans formulated without the use of SA/OR had been regarded as mainly political documents and were never implemented. The analysis in the plan was not a substitute for political judgment, but an aid to it. Analysis helped to identify feasible alternatives, pinpoint bottle· necks, etc. Some of the various planning committees remained intact and evolved into forums for the exchange of opinions on economic issues. This led to wide and open support of the plan itself and enriched the subsequent decision process. Finally, the plan document presented a consistent, concrete development strategy; explicit assumptions and consequences were spelled out. It was a result of interaction between intuition and analysis and between high-ranking officials and technicians. Because of these attributes and the drive for revitalization on the part of Korean leaders, the Second Five·Year Plan has come to symbolize a turning point in Korean history.

Systems Study of Agriculture in Mexico The preceding studies dealt with the decision problem at quite different levels of specificity. The first treated the case of a single resource (a fleet of motor buses). The second dealt with an investment project in which the project outputs contributed to three economic sectors: agriculture, energy, and transportation. The third was carried out at the level of a total economy, and the basic units of analysis were sectors, not commodities. The study reported in this section deals with a single sector in its entirety-<:rop agricul-

SYSTEMS ANALYSIS AND OPERATIONS RESEARCH

24

ture in Mexico-and the analytic building blocks are agricultural commodities and inputs. 6

Policy Applications

The Mexican Government and the World Bank collaborated for five years,

1970 through 1974, in the formulation and application of agricultural sector planning models. The work was carried out in three phases. The first, which continued to early 1972, focused on developing the basic methodology and demonstrating illustrative numerical results for the planning models. The Bank of Mexico was the primary sponsoring institution in this phase that culminated in the drafting of the agricultural chapters of the Goreaux-Manne volume published in 1973.' The agricultural model was named CHAC, after the Mayan rain god; the name became a shorthand description of the entire project. At the conclusion of this demonstration phase, the Mexican Government began the second phase by broadening the scope of the work. The government invited one of the participating World Bank staff members to live in Mexico for two years and work in the Ministry of the Presidency to continue development of the methods and assist in applying these methods to Mexican agricultural policy questions. During this second phase, the model analyses served as the basis for a number of governmental policy papers on specific issues and for a more general document, which has helped guide policy formulation in the Echeverria administration (1970-76). In the third phase, the main thrust of the work has been carried on by a Mexican group working in the Ministry of the Presidency, with some continuing collaboration by the World Bank. CHAC and associated submodels have been functioning on Mexican Government and World Bank computers, as of this writing, for about three years. At the invitation of the Food and Agriculture Organization (FAD) in Rome, a joint project is being undertaken by FAD staff, the World Bank, and analysts from participating countries to 6 Some of the key references are as follows: The seven agricultural chapters, by L. M. Bossoco, J. H. Duley, R. D. Norton, and others in Multilevel PIDnning: Case Studies in Mexico, edited by Louis M. Goreux and Alan S. Manne. (Amsterdam: North-Holland Publishing Co., 1973) (Chosen by the Operations Research Society of America for the 1973 Lanchester Prize.) L. M. Bassoco and R. D. Norton, "A Quantitative Agricultural Planning Methodology," World Bank Staff Working Paper No. 180, May 1974. L. M. Bossoco, R. D. Norton, and J. S. Silos, "Appraisal of Irrigation Projects and Related Policies and Investment," forthcoming in Water Resources Research. 'Multi-Level PIDnning: Case Studies in Mexico, edited by Louis M. Goreux and Alan S. Manne. (Amsterdam: North-Holland and American Elsevier, 1973)

SA/OR IN THEORY AND PRAcrICE

25

make further extensions and applications of the method in other parts of the world.

The Problem and the Model

Agriculture often presents complex issues in which many elements are interrelated. Product-pricing policies, for example, can significantly affect employment rates, income distribution, and the foreign trade balance. Mexico, with its diversity of ecological regions and agricultural products, was faced with several complex agricultural policy issues. During 1968-1970, Mexican officials began to doubt whether the agricultural sector could continue to provide for domestic consumption needs as well as to generate most of the country's foreign exchange, as it had for the preceding two decades. They saw that it was becoming increasingly costly to open new arable lands and tap new irrigation supplies. Signs of a long-term slowdown in agricultural growth had been appearing, and officials were also concerned about the capacity of the agricultural sector to absorb a significant portion of each year's new entrants to the labor force in productive employment. They asked how fast the sector could grow, what the employment possibilities were, and how the answers to these questions would be affected by changes in factor- and product-pricing policies and associated changes in crop composition and productiontechnol~es.

The attempt to meet such issues and concerns led to the formulation of the model CHACo The model describes both the local production conditions for all major producing localities in Mexico and the sector-wide behavior in agricultural markets in terms of equilibrium price levels and quantities marketed domestically and internationally. On the production side, the model is based on microeconomic· farm production cost and input data, by crop and technique, for each locality. This feature makes it possible for agronomists and other field specialists to contribute directly to the specif"tcation and annual updating of the model's production relationships. On the market side, estimated consumer-demand functions and export-import and transportcost parameters are utilized; a linear programming algorithm· is used to guarantee the appropriate market-equilibrium points. Essentially, the production aspects of the model are based on agronomic information; economic concepts such as crop-supply functions are derived by solving the model, rather than serving as inputs to the model. In this respect, CHAC helps transform engineering data into economic relationships and illustrates an essential feature of SA/OR. Thirty-three products are included in the model, and their interdependence in both demand and supply is reflected. In most producing areas of

26

SYSTEMS ANALYSIS AND OPERATIONS RESEARCH

Mexico many alternative crops can be grown, so an increase in production incentives for one crop is likely to influence output levels of other crops. The linear programming format, which allows alternative crops and technologies to compete for the same limited endowments of land and other local resources, is quite appropriate for capturing this kind of interdependence. More detailed studies of local choices have been made by detaching and solving some of the regional-production submodels of CHAC as separate models. In many cases they were considerably refmed to address local problems more adequately. The format of CHAC and the submodels are detailed, yet flexible and easily altered for different kinds of analyses. For example, the use of land, irrigation water, and labor is represented on a monthly basis throughout the annual cropping cycle, but for some studies water use has been specified fortnightly; for others, annually, depending on whether irrigation management is of concern. Several distinctions are made in the representation of labor, according to whether it is provided by farm-family members or by hired workers, and according to regional differences in wage rates. With this kind of structure, CHAC's solutions simultaneously describe aggregate characteristics of agriculture-such as total output levels, prices, producer-income levels, and foreign trade-and more concrete output details considering such variables as crop, technology, and region (and the interrelationships among them), including monthly employment figures in each region. CHAC is an optimization model in the mathematical sense. However, it is structured so that the optimization serves only to guarantee simulation of certain types of market equilibria, e.g., perfect competition or monopoly. In this context, alternative policy instruments, or packages of instruments, were put in the model to simulate the sector's likely responses. In this way, the probable consequences of each package could be outlined in terms of many variables, such as employment, sector output, income levels, regional distribution of income price levels, exports and imports, and use levels of factors other than labor. The model, therefore, was defmitely not used to devise an optimal program, but to assist policymakers by estimating the numerous consequences of many alternative, possible programs.

Specific Applications In making applications, the models were first managed by a working group in the Ministry of the Presidency and later jointly operated by this ministry and a newly established Agricultural Sector Coordinating Commission. The applications of CHAC and its submodels primarily concerned the following areas:

SA/OR IN THEORY AND PRACTICE

27

• Overall sectoral strategies, in terms of output, employment, foreign trade, and associated investment requirements; • Pricing policies for com, wheat,and other crops; • Factor pricing, particularly for agricultural machinery and water; • Export strategies to take account of comparative advantage rankings among crops; and • Project appraisal for irrigation works. Some model results played a significant role in the policy-discussion process; some examples are worth mentioning. It was estimated that the GNP growth target of 8 percent (in real terms) implied an agricultural growth rate in excess of 5 percent. Corresponding estimates were made of the necessary rates of increase in arable land and irrigation supplies. After adjusting for likely rates of rural-urban migration, it was also shown that this growth rate alone would not absorb all new entrants into the rural labor force. Complementary measures to create employment were outlined with the model. These involved: • Raising the price of agricultural machinery relative to labor, thus slowing the rate of displacement of field labor; • Creating incentives for promoting the export of labor-intensive crops; and • Making livestock-grazing operations more intensive to free land for growing crops, because crops provide more employment per hectare. Calculations were also made with CHAC of the level of comparative advantage by crop on international markets. It was demonstrated that to achieve Significant increases in agricultural exports it would be necessary to shift radically the types of crops exported. Finally, calculations of crop responsiveness to price changes and effects of changes in price on other crops helped to establish new price-support levels for food grains. Continuing methodological studies by the CHAC group involve the incorporation in models of risks faced by producers. They are also studying means of incorporating estimation of capital-labor substitution parameters and are studying consumer-demand structures in linear programtping, in methods of project appraisal, and in other related areas. The CHAC exercises led to many specific applications, but CHAC was probably equally useful in its more general results-helping to provide a focus for continuing policy discussions and indicating specific trade-offs and kinds of interdependence. It should be pointed out that a model could not have been used in this way without the well-trained personnel in high positions in the Mexican Government who were willing to sponsor and help interpret the work.

28

SYSTEMS ANALYSIS AND OPERATIONS RESEARCH

Blood Bank Inventory Control Returning to the scale of the fllSt example, this illustrates in more detail the process of SA/OR at the single-project level.a The problem addressed by the study is how best to distribute supplies of whole human blood for use by hospitals. Similar problems-how to make the best use of limited resourcescan be found on the agenda of any government department. The methods developed to solve the problem can be used whenever a necessary commodity is in short supply, is needed in many places, and can be traded among local warehouses. The study was conducted using as the example the hospital system of the Commonwealth of Massachusetts in the United States. Many hospitals in Massachusetts had their own blood banks, drawing from their own donors as well as from the statewide supply maintained by the Red Cross. The difficulty was that individual hospitals were too small to maintain an efficient blood bank. Because of variations in demand, the Red Cross found itself supplying large amounts of blood that were hoarded by individual hospitals until outdated (whole blood cannot be used after 21 days). At the same time, other hospitals ran out of a given type of blood. The suggestion was made that a more unified system of blood storage could reduce the amount that became outdated and the number of times a hospital ran out of a type of blood. Since human blood is essential to the treatment of many illnesses and injuries, many communities maintain blood banks for the collection, storage, and distribution of whole blood and its derivatives. In the United States, where more than five million pints are transfused each year, blood banks are organized into regional systems consisting of 20-200 hospital banks. Cooperation among the individual hospitals is very loose, but is essential for efficient operation of the total system. The several systems suffer from: • Simultaneous shortages and outdating of blood resulting from malmstribution and systemwide deficiency of supply; • Inability to supply sudden large demands; and • High operating costs for emergency shipments and losses through outdating. The blood-bank problem has the classic features of inventory-control problems in general, but in more acute form. Here the problem is not to minimize a single quantity but to balance as well as possible two mutually a For a full account see John B. Jennings, "Blood Bank Inventory Control," in Analysis of Public Systems, edited by A. U. Drake, R. L. Keeney, and P. M. Morse. (Cambridge: MIT Press, 1972)

SA/OR IN THEORY AND PRACfICE

29

opposing effects, shortage and outdating. The administrator of the blood bank must be made aware that both cannot be minimized at the same time and that he must decide the nature of the balance between them. The task of the SA/OR worker is to devise means to reduce both shortage and outdating, while maintaining the balance that has been decided on. This is a specific task; it does no good for the worker to start extending his analysis to questions of how more blood can be collected or how many persons should be getting more whole blood. These are other problems; if the specific task of reducing shortages and outdating can be accomplished, it will contribute to the solution of more general problems when they are attacked. It is seldom advisable to tackle the most general problem until one knows something about its parts. The following account of the steps taken to solve the problem illustrates the general methods and shows how SA/OR functions.

Definition of Objectives

It is desirable to take the following measures: • Minimize the delay in meeting requests for blood; • Ensure that the highest quality blood is supplied; • Minimize the cost of operating the blood·banking facilities; and • Minimize the total demand for blood by the blood bank (as opposed to the demand by physicians, which is an external constraint).

Approach to the Solution

Three major tasks were identified as being of importance to the solution of the problem: 1. Develop a conceptual, quantifiable model of the blood-bank operations of a single hospital. (In fact, the model depicted in Figure 1 was used.) Assigned blood is blood designated by control number-Le., blood contained in a designated bottle-for use by a doctor who has requested it. Unassigned blood is all other blood in the bank. In this model, the inputs of blood are in part controllable (orders placed in the central bank) and in part random and uncontrollable (donors and other hospitals). The output to physicians is altogether uncontrollable; the output to the outdating exit depends on the random inputs and outputs. Thus, the levels of stock, the shortages, and the outdates are random variables, and the behavior of the model is best studied by means of simulation.

30

SYSTEMS AN ALYSIS AND OPERATIONS RESEARCH

2. Study and compare the behavior of the model with the behavior of inputs, outputs, and stock levels in an actual hospital. Figure 2 shows the results of such a study. In this figure, the blood inventory operating curve represents the relationship between annual total shortages and outdating, both expressed as percentages of the annual total blood transfused. Each point on the curve corresponds to a different inventory-control policy, characterized by the reorder level S (the stock level at which one reorders a new supply). Each. represents simulated operations under alternative ordering conditions. Points between levels 12 and 18 are compared to actual hospital experience and found to agree very closely. The understanding of the system conveyed in Figure 2 allows one to select the optimum ordering level, depending on the respective utilities assigned to shortage and outdating, and to investigate the effects of changes, e.g., in the freshness of blood at input time. 3. Investigate the effects of coordinated control. It now becomes possible to study the operating curves of systems in which several hospitals pool their inventories in different specific ways. In fact, two distinct control methods were studied: the common-inventory method and the threshold-transfer method. In the first method, several hospitals pool their supplies and establish a common bank-common in the sense that anyone hospital may draw on the supply, even though all the participating hospitals share the storage responsibilities. This method involves much intercommunication and requires transportation. The second method involves dispersed banks and a transfer policy as follows: Whenever the inventory level for any specific type of blood, at anyone bank, falls below the level of one unit, a transfer is initiated.

Other hospitals Central blood bank Donor service Random donors

Physician ~~::::::;~!:req~u~es~t~s_-1

Unassigned blood inventory

Physician releases

Assigned blood inventory

Outdate FIGURE 1 Single hospital blood-bank inventory model.

Outdate

t - - - Usage

31

SA/OR IN THEORY AND PRACI'ICE

16

i= 12 z w

u a:

w

!:. ~

8

z

~
S = 10

0 I-

::>

0

Inventory operating curve

4

0'--

o

"'--

"'--

-'-12

-'16

....... 20

-'

24

SHORTAGE (PERCENT) FIGURE 2 Shortage-outdating operating curve: an independent hospital

• Each bank will lend only the units it has in excess of one unit. • When more than one bank is willing to lend, the banks are solicited in order of decreasing amount of stock (on the most recent inventory listing) until a willing lender is found. • The amount to be given per transfer is five units; if less is available, then as much as the lender can spare without reducing his stock below one unit. • Stocks are selected for transfer on the basis of age, oldest first. The results of a study of this particular policy are depicted in Figure 3. It was found that the threshold-transfer policy permitted a 54-percent reduction in both shortages and outdatings when five banks were pooled, and that it resulted in about a sixth as many interhospital shipments per year as the common-inventory policy. Jennings concludes his account with the following summary: 9 The accomplishments of this study are threefold: First, the whole blood inventory problem for a single hospital has been structured and 9 Ibid.

32

SYSTEMS ANALYSIS AND OPERATIONS RESEARCH

i=

z

w

U

a:

w

!!:. ~

z

~

« o ~

o

4

SHORTAGE (PERCENT)

FIGURE 3 Shortageo<>utdating operating cunes for the threshold transfer.

several classes of alternative policies identified. Second, a realistic model of the whole blood inventory system in a &rOUP of hospitals has been developed for use in the operations of such systems. Third, the feasibility of testing the effects of specific policies using computer simulation has been demonstrated, a number of such policies have been examined in detail, and the simulation programs have been documented for use in analyzing specific systems. The analyses performed should provide useful insights with direct applicability in a variety of blood banking systems. Specifically, we have determined (I) the range of improvements in shortage and outdating performance that can be expected in systems of various sizes, (2) the nature of the corresponding support system capabilities that must be provided, and (3) the fact that a group of only five cooperating blood banks can achieve most of the gains available in larger coordinated systems. The multibank system study was turned over to the Massachusetts Red Cross. It has been operating now for three years and the improvements predicted have been borne out in practice.

Brief Examples Three of the preceding examples illustrated the use of SA/OR as an aid to the realization of the planning process. Other examples of this kind are briefly described and categorized in Table 1. The table lists the area of application,

SA/OR IN THEORY AND PRACfICE

33

the prototype problems, the scope of specific studies referenced, the objectives, the techniques, and the known extent of implementation. The items are keyed to books and articles from which they were drawn, listed in Appendix B. These readings of SA/OR practice are relevant to the problems of developing countries. It will be noted that some of the studies listed in Table 1 were not implemented; there are many reasons for deciding not to implement the findings of an SA/OR study. For instance, an exploratory study may be directed toward a policy currently forbidden by law, as in the case of airline overbookings. There may be a shift of policy or, more simply, an external reason for not implementing the recommendations arising from a study of deliberately limited scope. The techniques are heavily weighted on the side of programming modelsi.e., linear programming in the original sense, nonlinear programming, mixedinteger programming-and on the side of econometric models, usually also linear. This is perhaps to be expected when one is dealing with the aggregations that are the proper concern of national and regional plans. There is another side of SA/OR, however, that is highly relevant to the goals of the LDCs. This is the tactical work of increasing the effectiveness of all the organizations at work in the economy. Where the planners are concerned with making the most rational use of things as they are, the SA/OR practitioners are concerned with changing the production functions that enter the macro-level planning equations. Table 1 also lists examples of studies devoted to this purpose. They are of the general type of the blood-bank studythat is, they involve a system of narrow scope that is relatively easily observed and that admits a relatively simple measure of effectiveness. Further material can be found in the abstracts contained in International Abstracts in Operations Research, published by the International Federation of Operations Research Societies (IFORS).lO

lOElsevier North-Holland, P. O. Box 211, Amsterdam, The Netherlands.

TABLE 1 Examples of the Use of SA/ORt Techniques Used

Results Implemented

Evaluating alternative ways of meeting increasing demand for a product

Decision tree

Yes

A model of the hide-leather sequence in a tannery

Optimizing hide-leather sequence of tannery process

Linear programming

Yes

Maintenance simulation

To provide a routine material usage forecast and a means of evaluating material usage policies

Simulation

Yes

Investment in electric power industry

Optimum investment decision making

Linear programming

Yes

Electric power system expansion

To determine the least-eost mix of capacity between hydro, nuclear and thermal, the size of the plants to add to the system, and the timing of these additions

Dynamic programming

Yes

Area

Prototype Problems

Specific Studies

Objectives

Industry

Production planning and scheduling; Distribution systerns; Site selection; Capacity expansion; Inventory control; New products: planning; sequencing, replacement, maintenance

Decision model for capacity expansion

IN ,J>.

Energy

Technology selection; Investment planning and scheduling; Capacity expansion

Transportation Highway-networks, bus scheduling, truck routing, automated urban intersection traffic control, airline scheduling, modal split, resource allocation, communication

Computer control for urban intersection

To develop computer control scheme that will provide traffic responsive correction action for a critical intersection in urban system

Computer simulation

No

An airline overbooking model

To determine optimum airline booking policy

Markov process, dynamic programming

No

Water Resources

River basin utilization improvement

To determine optimal water resource allocation in a proposed water conservancy district

Linear programming

Yes

Assessment of rural water supply program effectiveness

To improve the methodology of impact evaluation by determining the role of resource constraints and complementary inputs in the development of the rural water supply program in Thailand

Feedbackcontrol, systerns theory

Yes

Investment, evaluation in irrigation

Evaluation of investments in irriStochastic gation, considering complex hydro- programming, logic and economic interaction chance constrained programming

Unsure

Irrigation feasibility study

Determination of irrigation systern which would best serve local community and the nation as a whole

Unsure

Planning, design and operation of reservoirs; River basin development and management; Water quality management

IN VI

Irrigation

Irrigation water supply; Investment decision making; Technology selection; Irrigation quantity and frequency

Mixed integer programming

TABLE 1 (Continued) Techniques Used

Results Implemented

Optimizing size of exploited natural animal population

Stochastic programming; Inventory theory

No

Modification of poultry processing to minimize water flow through the plant

Linear programming, network theory

Yes

Emergency admis- To design and institute a presions and the pre- discharge ward suitable for all discharge ward adult patients of both sexes and all specialties during the last few days of their hospital stay

Stochastic processes

Yes

Simulated queueing approach to scheduling in surgery

Queueing theory, simulation

No

Area

Prototype Problems

Specific Studies

Objectives

Agriculture

Crop planning; Natural animal population and exploitation; Poultry; Pest control; Equipment use planning

Optimal animal population model

Water management in poultry processing

~

0\

Health Care Delivery

Manpower-facility planning, patient-staff scheduling, hospital blood bank inventory, menu planning, diagnosis-patient treatment, maintenance, resource allocation, emergency room, multiphasic screening

Simulation of several surgical scheduling procedures, for use in designing facility size and selection of optimum scheduling policy

Urban

Planning and Management

Education

IN .....:l

Urban water and land resources management; Community development; Emergency units development and allocation; Refuse collection; Waste management

New community development process

To determine land development and allocation process and analyze fmancially

Linear programming

Yes

Urban public decision making

Integrating the interacting components of urban system

Simulation

Yes

Planning and investment in education; Information systems; Allocating faculty; Organizational structure; Library and housing system; Local school district management; Classroom utilization

Information system design for educational management

Design and implementation of a student information system for a school of engineering in a university environment

Management information system

No

Educational plaJining

Predicting pupil enrollment in an education system

Markov process

No

Tourism

Investment; Resource allocation; Management

Investment alloca- Investment allocation for tion model for touristic projects tourism sector of a developing country

0-1 Integer programming

Unsure

Criminal Justice System

Courts scheduling; Patrol scheduling; Crime information system; Prison system effectiveness

Analysis of a total criminal justice system

Linear models, feedback dynamics

No

tFor references, see Appendix B.

Describing means of modeling the criminal justice system

3 The Application of SA/OR

OVERVIEW In the preceding chapter, SA/OR was shown to be applicable to the operating and resource-allocation problems of organizations. The use covered problems of planning, control, and effectiveness, and was broad enough. to cover organizations ranging from newspaper stands to nations. The scientific method gains its power from its ability to abstract, to generalize, and to quantify; the systems approach gains its power from its ability to consider the whole in relation to the parts and in relation to the environment. SA/OR is not a substitute for administrative judgment, but an aid. It is not a substitute for intuition; rather, it helps to channel it. It is an applied science that contains a research component, a push into areas where knowledge is not sufficiently elaborated to allow valid and universal rules to be formulated and applied. Sagasti,u however, cautions: ... SA/OR, contrary to what most people would like it to be, is not a value-free process. The definition of objectives and objective functions, the specification of what constitutes the "best," the "appropriate," or the "most convenient" result of a particular operation are clearly not value-free decisions. Here we have the problem of conflict between individual rationality (at the enterprise or government department level) and social or collective rationality. Often it is not possible to assume that both 11 Francisco

Sagasti, private communication to the Panel.

38

THE APPLICATION OF SA/OR

39

of them coincide, and we have many cases where the SA/OR worker helping in the pursuit of individual objectives might act against the social interest.... If the process deals with problems, it begins with the recognition that a problem exists. Recognition is a function shared between the administrator, who uses the services of the group, and the group itself, which helps the administrator. The administrator may perceive an area in which he wishes he had better understanding or better control. The research group may suggest an exploration or the use of a new technique; it may call attention to the apparent failure of an existing process. Many successes of SA/OR arose from merely a fresh look at an old problem occasioned by an administrator's dissatisfaction with things as they were. • Research is a cyclic process. It proceeds from problem defmition through model formulation to data collection to tentative conclusions. These are tested by trial and the assumptions and defmition of the problem are reformulated if necessary. The cycle is repeated until a satisfactory approximation of reality is attained. This description is not strictly accurate, since the process is not linear. It does not go from one step to another, but progresses in all directions at all times, depending on the state of affairs from moment to moment. With this qualification in mind, one may use the linear description as a thread to be followed through a discussion of systems analysis and operations research.

THE PROCESS IN DETAIL

Defining a Problem The process begins with the recognition that a problem exists: an administrator or an authoritative group is dissatisfied with the state of things, or feels that the information is incomplete. Or perhaps it begins with a feeling on the part of the SA/OR director that an important system is not well understood and that attempts to control it are wrongly conceived. In the frrst example (chapter 2), the Baroda bus system was becoming overloaded. The Gujarat State Road Transport Corporation wished to explore the potential improvements to be gained by rescheduling, before putting on extra buses. In the second example, planning the water and power system of West Pakistan was not a primary goal, but a means to an end. The primary goal was to develop and apply an effective method for selecting projects and for assigning priorities so that satisfactory use would be made of development funds. In the third example, the problem arose from the dissatisfaction of the

40

SYSTEMS ANALYSIS AND OPERATIONS RESEARCH

Korean Government with the results of its planning process. In the fifth example, the problem of whole-blood inventories first appeared as administrative dissatisfaction with the results of the current system of distribution. The administrator felt that too much blood was wasted through outdating and too many requests for blood were not satisfied. It should be noted that this dissatisfaction was not focused on inventory control or on any other feature of the system, but rather on the accomplishments (the output) of the system compared to what was wanted. Other examples come readily to mind: fruit is arriving at its destination in poor condition; a clinic fmds that it has too many patients and seemingly not enough doctors; the conflicting demands of the economy for irrigation, power, and navigable water need reconciliation, or at least resolution. Whatever the origin of the problem, a preliminary definition is essential. This definition is arrived at by agreement between the administrator and the SA/OR director, with the understanding that their perception of the problem may change in the course of the work. At this same time, an agreement is reached on a reporting and conference schedule, at which the administrator and the director, with such members of the SA/OR team as are needed, will review progress, review their definitions, determine the value of further work, and revise their plans if necessary. The understanding that the scope and direction of the work may change is vital to the entire process. It is at this point, in the defmition of the problem, that the director must share the administrator's concern and understand his viewpoint, while adhering firmly to his own understanding of the way the problem will be attacked.

Finding a Model A model is a deliberately conceived symbolic abstraction from reality. An input-output model represents industries as abstract sources of supply and demand and represents the relationships among them by multiplicative factors (e.g., a million monetary units' worth of shoes requires SO thousand units' worth of transport, and a million units' worth of transport requires 80 thousand units' worth of shoes) that can be combined and manipulated by arithmetic and algebraic methods. A probabilistic model represents the occurrence of events that are subject to random and fluctuating forces. Aformaiized decision model consists of a set of (assumed) rules of behavior that vary with defined types of circumstance; such models are normally used in conjunction with probabilistic descriptions of the system and are combined into a simulation model. The use of models in SA/OR involves a complete and constant awareness

THE APPLICATION OF SA/OR

41

that the model is at best an approximate representation of reality. If the representation accounts for the main features of the system under study it is satisfactory, but it is not sacred. Figure 1 (po 30) is a symbolic representation (model) of the distribution of whole blood through a medical system. Each arrow represents a probabilistic or deterministic phenomenon, which in the numerical version of the model will be represented by a statistical distribution function or by a rule of behavior (e.g., "if the date on the blood container is more than 21 days earlier than the date of the inventory, discard the container and its contents"). Either phenomenon will take its part in a system of equations or in a program for computer simulation. In example 2, the behavior of an electric power system was modeled by simulation based on engineering principles, and the behavior of the irrigation system (considering such variables as rainfall, state of the water table, salinity, and usage) was modeled by a combination of probabilistic rules and rules derived from hydraulic engineering, meteorology, and physical chemistry. The economy was grossly modeled by a seven-sector representation. In example 3, the economy was represented by an input-output matrix. The factors in the matrix were developed by a combination of data analysis and expert judgment. They were related to one another, and to reality, by a process of repeated review, prediction, and modification. , The construction of manageable models is a principal activity of SA/OR. The administrator need not understand the process by which the model is developed, but the model itself should be altogether comprehensible. A part of the basic agreement between the administrator and the SA/OR group is that the group is entitled to its models as a craftsman is entitled to the design and use of his tools, but it is not entitled to a private mystique. In the same spirit, the administrator is not required to think in terms of the model con· cepts, nor to manipulate the symbols used by the SA/OR director. He is required, in his own interest, to meet the director halfway.

Data Collection It is often suggested that lack of data will be a significant, and perhaps insurmountable, obstacle to the success of SA/OR. Most SA/OR workers would disagree with this view for two reasons. First, if data are lacking, the system under contemplation has not been really examined, and the chances are that the collection of a small or moderate amount of data will yield valuable results. That is, the lack of data is a sign that data collection will payoff. Second, SA/OR is an experimental science, and the SA/OR group will need

42

SYSTEMS ANALYSIS AND OPERATIONS RESEARCH

to obtain its own data in its own way. Historical statistics, compiled for other purposes, are often unsuited to SA/OR needs. Instead of using such statistics, the group must make direct contact with the elements of the system. For example, it may be necessary to talk with workers, foremen, planners, and executives, and to see the land, observe the operations, and make relevant measurements. The administrator, not understanding this basic need for firsthand information, may view the presence or absence of historical data as a vital factor in planning a study, or even in deciding whether the study is feasible. This is a mistaken view.

Preliminary Conclusions and Test Design If the data have somehow been collected and processed through whatever numerical calculations the model demands, some tentative conclusions can be drawn. They must be formulated in administrative language and discussed with the administrator within his own framework. In example I, for instance, the proposed changes in the routes and timetables must be thoroughly discussed with the executives of the bus system to arrive at a solution that is operationally acceptable. Out of these discussions a changed program will emerge for test~ng by the simulation process. Several rounds of such discussion may be needed, during which the SA/OR group thoroughly explains the bases for its conclusions and works with the executives to develop the program that is fmally implemented. In the fifth example, the conclusion might be that a group scheme should be set up in which subsystems of five hospitals form common blood-delivery systems, fed by a central bank, with local interchange. This conclusion is subject to test, because it has been reached by calculation and by reasoning based on abstractions. Nevertheless, it is a conclusion to be discussed, examined for its implication in real life, and subjected to scrutiny by the administrator for its viability. The preliminary conclusions carry questions with them, some vital and some that, after examination, turn out to be nonessential. The source of supply of blood, its availability to different segments of the population, and its connection with state policy are factors that go beyond the initial equations of the SA/OR worker and that must be dealt with by the administrator. If the system is to be tested, which group of hospitals can be picked for the test? How can the test be put into a suitable framework so that it will be accepted by the medical profession? By the people? These questions cannot be answered mechanically, within the framework of the model, but they should not be answered in a totally political framework. The test conditions must be devised so that the maximum amount of information can be ob-

THE APPLICATION OF SA/OR

43

tained, subject to any necessary political constraints. It will be necessary for the administrator and the director to work together in setting up test conditions. This prescription applies almost universally; only seldom can important innovations be tested without interfering in some way with the course of current affairs.

Implementation All of the research and testing discussed in the preceding sections will lead to a practical program of action. This program will evolve through extended interaction of the SA/OR group with the administrator and his executives; in the course of these conferences and discussions the factors making up the problem and their proposed treatment will be examined. On the basis of the understandings reached by this process, plans for implementation of a practical program are developed. The SA/OR group will normally have to take an active part in the implementation of the program. There are three good reasons for this: I. In the early stages of the implementation new factors may arise, factors not discovered in the testing and development phases, and therefore unforeseen. The SA/OR group is best qualified to assimilate the action of these factors into the model and suggest ways for coping with them. 2. The operating executives responsible for implementation of the program will have to make day-to-day interpretations of the basic principles on which the program is founded, and the SA/OR group can assist in the interpretation. 3. Active participation in the implementation of their recommendations is an essential element in the training and development of the SA/OR workers themselves. Of course, none of this means that the SA/OR worker is expected to become a line executive while he is a member of the research group.

Redefinition of the Problem The first round of work will have brought more light into the problem area and will usually justify a refmement or change of the problem definition. For example, a manufacturing plant may fmd that it can operate more efficiently if its day-to-day mix of products is allowed to depend on local conditions rather than on a planned schedule made at corporate headquarters. The

44

SYSTEMS ANALYSIS AND OPERATIONS RESEARCH

problem then becomes one of obtaining freedom from corporate control; this means broadening the study to take into account factors initially treated as part of the environment. For instance, a study of bus scheduling may show that the city is developing in the wrong direction and that what is needed is a study of factory location and housing development. Or a study of fertilizer needs may reveal a crucial lack of agricultural credit facilities, that is, the fertilizer cannot be used under the existing financial conditions of the farmers. More generally, it is a truism of SA/OR that a systems study will show which factors in the environment of the system are acting as critical constraints that keep the system from operating freely. Frequent conferences for problem redefmition are needed between the administrator and the director if maximum value is to be obtained from an SA/OR program.

Use of Electronic Computers It is sometimes felt that SA/OR is synonymous with the use of giant computers. The large computer is an important tool of SA/OR, and for some applications it is indispensable. But much of the work can be done without recourse to computers. Moreover, when a large computer is required, it need not be located within the SA/OR group's center of operation. The use of distant computers is practicable, either by mailing the data to the computer center for processing, or by the use of telecommunications. Also, the major computer manufacturers, and many service agencies, maintain standard programs for the solution of a number of SA/OR problems. The SA/OR group should, of course, have at least one member who is thoroughly familiar with the use of computers and can write programs when necessary. Immediate access to a small- or medium-sized computer for dayto-day work is also desirable. Computers of modest cost-perhaps no more than a few thousand dollars-can be extremely useful. Among the many uses of computers, two are outstanding: the solution of systems containing many mathematical equations and the simulation of systems models in action. Systems with many equations arise in connection with linear programming methods, in economic studies involving the use of inputoutput tables, and in other macroeconometric techniques. Simulations arise in many contexts; generally it is necessary to run a simulation when there is no obtainable mathematical solution for the equations that describe the system.

THE APPLICAnON OF SA/OR

45

Role of Experiment In most cases, field experiments are both desirable and necessary, either as a check on conclusions derived from theory, or as a way to get information that can be obtained in no other way. Although it is true that simulations can yield valuable information, they are the product of assumptions about system behavior that are imposed by the SA/OR worker, whose perception of the behavior of the real world may be mistaken. An experiment need not necessarily be run on a large scale; even piecemeal experiments can be of great use. Experiments should be planned with great care, after a preliminary study of the predictions made by analysis of the models and by simulated experiments. Full-scale experiments are bound to be costly, and they may have undesirable side effects, ranging from worker dissatisfaction to large economic losses. It is therefore desirable to make a preliminary analysis of the possible outcomes, and of the conclusions that would result, to see whether the experiment is necessary and to determine how to get the maximum information from it.

4 Organizing for SA/OR

There are several ways to organize SA/OR activity. This chapter discusses the factors that should be considered and suggests some forms of organization that can be used. There are universal requirements, common to all organizational forms, with respect to the administrator, the director of SA/OR, and the SA/OR staff. Except for these constraints, a reasonable amount of flexibility is possible. A form that has been found particularly successful consists of a relatively small core group. Specialists and experts work with the group on an ad hoc basis to provide skills and experience required by specific projects.

KEY PERSONNEL The prime requirement for the successful use of SA/OR is that the administrator (government executive, manager of a business, etc.) see the need for it and be willing to consider the recommendations seriously when he makes his decisions. He must also be prepared to provide assistance in the gathering of data. He must be thoroughly familiar with the operation under study and with the customs of the personnel involved. He must also have the authority to see that the needed data can be gathered and to put into practice the solution arrived at by the SA/OR team and approved by him. The director of SA/OR should have administrative rank equal to that of 46

ORGANIZING FOR SA/OR

47

any other principal advisor to the administrator. This is a double-edged requirement: the person chosen as director must have personal stature appropriate to his position as a principal advisor, and he must be afforded ready access to the administrator. The director must have the administrator's confidence and trust. He should be capable of maintaining a confidential relationship and of creating a corresponding attitude among his staff. The core staff should be selected for a balance of skills. The professional staff must include an applied mathematician and an expert in computer usage, and all its members should have some training in science. Although it is possible to allow some members of the staff to function as internal con· sultants with little external contact, it is most undesirable. All the staff should be prepared to work in all aspects of the group's program. It is not possible for a small group, or even a medium-sized group, to possess skills in all branches of engineering and science. The group must expect to recruit specialists and experts for particular projects, and should be able to work with them easily. Some experts will have scientific training; others may have primarily practical experience. It is essential that the group not engage in scientific snobbery that makes working with others difficult.

ORGAN IZATIONAL FORMS The organizational relationships into which the SA/OR activity must fit

will vary widely from one country to another. The follOWing sections discuss a variety of forms that may be used, without prescribing anyone. The fundamental requirement is that SA/OR workers be actively involved in problems important to the welfare of the nation, and that they not be confined to a back room.

Independent Groups In the industrialized countries, the general tendency is for each government agency either to set up its own SA/OR group or to contract with private firms or universities for SA/OR services. When the reservoir of skilled SA/OR personnel is small, the use of many small groups may be wasteful, or even impossible. It does, however, offer certain advantages: the dedication of the group to a single interest and the buildup of experience with the problems of a single agency. Since most SA/OR work is innovation, some way should be found to renew the outlook of a permanently dedicated group. This can be done, as in

48

SYSTEMS ANALYSIS AND OPERATIONS RESEARCH

the United Kingdom, by developing SA/OR as a career civil service occupation. Trained staff are rotated among the several groups that have been created.

A Central Group In the formative period, before a stock of experienced SA/OR practitioners has been accumulated, the most efficient form may be the single group, serving the entire government. This group can provide the general skills of SA/OR, manage the necessary computing facilities, and recruit specialists whenever they are needed for specific projects. The central group is easier to support, on a long-term basis, and offers the individual staff member a more attractive career potential. The work will be prevented from deteriorating by varying the assignments. Service to the several agencies can be provided either by temporary postings, amounting to a transfer of staff, or by assignment of a team without changing the team's location. Since SA/OR involves intimate cooperation between research workers and consumers of research results, transferring the work site to the premises of the consumer is usually desirable.

University Consulting In some cases, the SA/OR group may appropriately be set up under the auspices of a university. This is advantageous when training new professional SA/OR workers, who can combine academic training with practical experience. The faculty, too, profits from combining theory with practice. A danger is that the ends of education are likely, and rightly so, to take priority over the needs of the client. On the other hand, a university is a center of expertise, and a university-based SA/OR group will find it fairly easy to recruit any expert needed to fill out a project team. Many universities in the industrially advanced countries hav~ set up SA/OR groups, with excellent results.

Private Consulting Groups SA/OR is practiced as a professional activity for professional fees by many private firms. There are advantages and disadvantages. The administrator may find it advantageous at the beginning of his experience with SA/OR to use the services of a professional group, thereby making

ORGANIZING FOR SA/OR

49

no commitment to the existence of a group on his own staff. The private firm can usually provide a wider range of skills than the administrator could command within a group of his own, at least in the early days. The disadvantage, from the viewpoint of the LOCs, is that a small private consulting firm tends to lead a precarious existence, alternately overcommitted and underfed. Disappointing experiences of this kind may hamper the development of SA/OR by turning good potential practitioners away from it. Another disadvantage is that a consulting team is usually hired to 'address a particular problem, then is often not available at the time the fmdings of its research are put into practice, when its knowledge can make the difference between a successful and unsuccessful project. Though using a consulting team is an easy way to begin the application of SA/OR, foreign consultants can never provide the continuity and concentration on local problems that can be supplied by a local group.

LOCATION OF THE GROUP Since SA/OR exists only if someone wants to use it and can support it, the simplest answer to the question of location is that an SA/OR group should be located wherever it is wanted. Nevertheless, brief discussion of some factors to be considered may be helpful. The names of locations are indicative only, since formal designations vary from country to country.

Chief Executive's Office It is quite possible that the chief executive will wish to establish an SA/OR activity for his own use. However, the pressures are great, the time span of decision making tends to be short, and the scope (}f the scientific method tends to be constricted. The chief executive may prefer to appoint a scientific advisor who will keep in touch with SA/OR activities, raise questions from time to time, and initiate projects of national interest.

The National Planning Agency A group in the national planning agency can be of great value. The application of such techniques as linear programming, input-output analysis, and macroeconometric modeling can assist materially in the rationalization of the planning process.

50

SYSTEMS ANALYSIS AND OPERATIONS RESEARCH

A Ministry In the industrially advanced countries experience shows that SA/OR can be profitably used by single ministries. For example, the ministry of health can benefit from studies of the efficiency and effectiveness of hospitals, of rural health services, of paramedical programs, etc. The ministry of transport can use studies of the efficiency of existing road networks, of transport systems (whether nationalized or not), and of the transport needs of the country. The ministry of education can benefit from studies of the optimum location, size, and curriculum of schools and colleges. All such studies, although not the exclusive province of SA/OR, can receive helpful contributions from its methods.

An Urban Government Urban studies relating to traffic control, location of service centers (such as fire stations, schools, police stations), management of purchasing and disbursing operations, and many similar problems can be successfully carried out by SA/OR methods. Urban governments are more likely to use consulting services (government, university, and private) than they are to support their own groups because their budgets do not normally include provisions for expenditures of the magnitude required to sustain an ongoing SA/OR ac·tivity. However, very large cities (i.e., New York and London) have their own SA/OR groups.

Subordinate Government Agencies Port authorities, fire departments, housing agencies, and railroads parallel urban governments in the immediacy of their problems and the probability that they will employ a diversified group rather than sponsor a group of their own.

Private Industry Many industrial firms have formed SA/OR groups, which they maintain because the benefits accruing from the increased effectiveness and efficiency outweigh the costs of maintaining the group.

ORGANIZING FOR SA/OR

51

EQUIPMENT AND FACILITIES On the whole, the facilities required b}S an SA/OR group are modest: work space, conference rooms, individual offices with blackboards, secretarial assistance, library access, transport, and other logistic support. Access to a large computer is essential, but as noted earlier, the SA/OR group need not own the computer. However, if the group does own a computer, the cost may be reduced by providing computer services to other government agencies. Otherwise, contractual arrangements must be made so the group can get computing time whenever needed. If there is no large computer in the country, the group may need a budget for travel and computer rental so that computation can be done elsewhere. The use of remote terminals should be considered. (Note that in example 3, the South Korean FiveYear Plan, the large-scale computation was done in Japan.) Much valuable computation can be done, however, with machines of modest cost. Within the past two years, small hand-held machines, the socalled electronic slide-rules, have appeared on the market at a cost of $100-$200. These machines are suited to exploratory work and to the manipulations of numbers incidental to the study of initial models. At a somewhat higher price, but still in the low thousands of dollars, fully prograrmnable machines with substantial memory capacities are available. Certainly there is no need to postpone the use of SA/OR solely on the grounds that computing capability is not available.

AN INITIAL WORK PROGRAM The resources of LDCs vary widely, so that it is impossible to suggest a uniform work program for all. Nevertheless, some general observations can be made.

Size of the Project It is important to recognize that the utilization of SNOR has been delayed as often by overoptimism as by hesitation. It is usually unwise to start out a new expert or team of experts on a large problem before their capability is tested on a more limited and less costly task. Before the larger task is begun, the client needs to learn whether the team has the flexibility that comes from experience or is blindly wedded to a single technique that may not always be appropriate. The team and the authority must learn to work together.

52

SYSTEMS ANALYSIS AND OPERATIONS RESEARCH

It is quite possible, and indeed useful, to embark on a large-scale project, as typified by the second, third, and fourth examples. However, this report addresses the development of a permanent, indigenous SA/OR capability based on local resources. This requires building a good foundation. At first, people with a theoretical background must be brought into close contact with operating problems, and practical people must be brought into close contact with the uses of theory. Many, or perhaps all, the obstacles mentioned earlier, will be present. In the beginning, it is advisable to work on a project that is at once of significant interest and modest size. If the project is too large, there will be delays in producing results and the danger of lack of realism (i.e., simplifications will be necessary to such an extent that the models will fail to represent reality adequately). In a small project, the administrator can stay abreast of developments and take an active part in decisions. His staff can become actively involved in the work itself, thus learning the uses of SA/OR and becoming accustomed to cooperative work with SA/OR professionals. Once the foundation is properly laid, expansion can take place as rapidly as circumstances permit. Normally, one would expect substantial growth to begin in the flISt or second year of work.

Kinds of Problems An extensive body of theory has been built up during the past 30 years and is ready for application with suitable adjustment of the parameters to the problem under consideration. Problems of inventory control, production scheduling, fleet management, hospitals, fire stations, police stations, libraries, banks, and many other problems have been thoroughly studied. The payoff to the administrator in increased efficiency of the operations under his control can be very high in terms of the cost of SA/OR effort. Problems of this kind are good for the initial stages. In the long run, the SA/OR groups should want and be able to deal with larger problems such as those connected with the balance of payments, management of raw-material resources in relation to the world market, and nationalland-use programs. These can even be dealt with at the outset, but there are risks in tackling such large projects too early.

Ad Hoc Teams There is need for cooperation among team members with differing professional backgrounds. The ability of the director and the members of the core

ORGANIZING FOR SA/OR

53

team to elicit cooperation from other professionals will increase with experience. The initial difficulties should not be underestimated. It is also essential that good working relations be established between the SA/OR staff and the members of the administrator's staff assigned to work with them. The SA/OR workers should make a strong effort to involve their counterparts in all phases of the work and should avoid any hint that they believe they are superior. It is easy for the SA/OR staff to recognize shortcomings in the operation being studied, and difficult for them to avoid feeling that if they were in charge things would be done better. In most cases, they would not be done better, but worse; new SA/OR staff should understand this notion thoroughly.

SOME CAVEATS SA/OR is an experimental science. The working team must acquire the habit of making its own observations without relying on the reports of others. This means getting out to the site of operations and observing the process in action. Any staff worker who feels this aspect of his work is demeaning should be reeducated or transferred. If SA/OR is seen as an elitist pursuit, conferring distinction and freedom from routine work on its practitioners, the work will certainly not be well done, and there is high probability that the group will perish. The models used to descnbe a system should come from observation, rather than a standard list. The initial reconnaissance should be made without preconceptions and should include discussions with the participants at all levels. Only after such reconnaissance should the SA/OR worker begin to formulate his model, select the important factors, and postulate the relationships that exist among them. The scope of a study should never be determined to fit a favorite model; an SA/OR team looking for a problem to fit its pet model is almost certain to fail. Sagasti 12 notes: [There are] several pitfalls involved in the use of management sciences which include: escapism in figures; clouding the issues to make them unintelligible; giving a "scientific" aura to pre-defined policies; wasting time and effort building useless models; gathering vast amounts of useless data; model fetishism; the use of a cannon to kill a fly; the fallacy of misplaced concreteness; and the satisfaction of ego trips of foreign and local researchers.... In general, the models employed should be as simple as the circumstances 12 lb id.

54

SYSTEMS ANALYSIS AND OPERATIONS RESEARCH

permit. If a simple model will explain 90 percent of the behavior of the system and can be handled mathematically without much difficulty, it is better than a more complex model that reflects the system more faithfully but requires extensive mathematical treatment and could delay the completion of the study beyond the date when it will be needed by the administrator. It should again be stressed that it is not the province of the SA/OR team to decide, unilaterally, what is the "best solution." The task is, rather, to work closely with the appropriate authorities to plan what is desirable and also feasible in the specific circumstances, taking into account local customs, environments, and supplies-not to try to apply some generalized "quick-fix" reputed to work elsewhere. For this reason it is important that as large a fraction as possible of the team be persons thoroughly familiar with local circumstances and that every effort be made to train LDC nationals, so that each LDC can have its own teams, working on its own problems, as soon as possible. In searching for useful SA/OR projects, it is wise to keep in mind a statement by P. M. S. Blackett: \3 Any organization that regularly varies its methods and observes the results will have come close to an optimal operation. If either of these elements, variation or observation, is missing it is likely that large improvements can be made by restoring it. This precept is useful to bear in mind during the course of the preliminary reconnaissance; it will help the SA/OR worker save time and will reveal areas in which relatively quick results can be obtained.

Failure Modes The method may fail, in a particular application, in a number of different ways. The following list is a catalogue of the failure modes considered by the panel; although it may not be exhaustive, it does cover those instances of failure that the panel members are aware of. The failure modes are not necessarily mutually exclusive. 1. The SA/OR workers may be merely incompetent-either sloppy in their work or too rigid in their thinking. 2. There may be a communications failure: the administrator expects too much or will not take the time to understand the scope of the recommenda 13 p . M. S. Blackett, "Operational Research," The Advancement ofScience (Apr. 1948):26-28.

ORGANIZING FOR SA/OR

55

tions made by the SA/OR group. (Obviously, charlatanism on the part of the .SA/OR people may mislead the administrator into expecting too much; but charlatanism is not the only source of great expectations-the group, or the administrator, may in all honesty simply overestimate their own competence.) Aside from the above, which represent human failure in a gross sense, there are further ways in which a competent group may fail to create an applicable set of recommendations. (Competence here means competence in the ordinary, realizable sense of the word; it does not mean that there will be no failures revealed by hindSight.) 3. The model chosen to represent the system may be erroneous, so that it does not fit the real world well enough. a. The model may be static, and so fail to reflect time-related changes that are determined either by outside forces or by processes inside the real system. b. Some of the parameters or system variables may be in fact unquantifiable. 4. The syst~m boundaries may be poorly chosen. a. The system may be so narrowly defmed that it cannot be adequately insulated from external forces, or that the SA/OR group neglects important control variables or what in medicine is called "side effects" and in economics "externalities." b. The system may be so broadly defmed that it cannot be adequately studied within the constraints of time and money imposed on the SA/OR group. s. The group's recommendations may be unrealizable. a. The system elements may resist control in one way or another (refusal to cooperate in execution of a plan; evolution of new phenomena). b. There may be trouble in the control system, inducing undesirable phase-shift phenomena, as when the time between observation of an output variation and the initiation of countermeasures is too long due to uncontrollable factors. c. It may be impossible to insulate the system from the environment, so that random forces intrude. 6. The data on which the recommendations are based may be bad. a. There may be significant errors or bias in reporting. b. The data may measure something other than it is believed to measure. 7. The parameter space may be too large. a. It may be impossible to get enough data to make meaningful estimates of the parameters. Then the predictions are arbitrary. b. It may be impossible to simulate the performance of the system over a large enough sample of points in the parameter space.

56

SYSTEMS ANALYSIS AND OPERATIONS RESEARCH

8. It may be that an SA/OR study is too costly compared to the potential value of its recommendations. a. The system may be already operating near its optimum point, in the absence of the study. b. The optimum may be discoverable by intuition.

If the likelihood is high that one or more of the failure modes will be activated, the proposed application is risky and the administrator and the SNOR group will want to consider its desirability very carefully.

5 Training SA/OR Workers

It has been noted earlier that the key elements in the success of SA/OR are the administrator who sponsors the work, the director who manages it, and the staff members who do it. The administrator needs to understand what SA/OR can and cannot do for him, and how best to use it. The director needs to know how to formulate projects, how to communicate with the administrator, with the staff, and with his colleagues, and how to manage a going operation. The staff, collectively, must be able to collaborate with their counterparts, must be competent in the applications of mathematics, and must have a talent for systems analysis. Training courses, conferences, and participation in actual projects can help to develop and reinforce these abilities.

TRAINING THE ADMINISTRATOR It is not necessary that the administrator be able to perform SA/OR work, and it is unlikely that he would want to. But he should understand what SA/OR does and how it can aid his work, so he does have something to learn. A list of the usual resistances to SA/OR appeared earlier in this report. They are summarized as follows:

57

58

SYSTEMS ANALYSIS AND OPERATIONS RESEARCH

• • • • • •

Change is impossible. Change is too difficult. Change is unnecessary. There are no data to be used as a basis for decisions. The risk is too great. SA/OR cannot help.

These opinions are held by many people, but they are often rationalizations of a resistance to change. It is unrealistic to expect to overcome all these resistances. It is sufficient for the introduction and strengthening of SA/OR if a few administrators in powerful positions succeed in overcoming them. Conferences, popular lectures, publicity of various sorts, and one- or two-day training courses for selected administrators are useful techniques for generating support.

TRAINING THE DIRECTOR The director of an SA/OR activity is an executive within his own sphere of action. He must have organizational and interpersonal skills and must be able to attract and lead creative staff members. He must also know how to work with administrators and executives, how to help them defme their problems, and how to help them put his recommendations into practice. Useful training can be offered through SA/OR fellowships carrying the option of studying abroad for one or two years. Fellowships can be jointly supported by the home government and by donor agencies. The fellow will combine postgraduate work at a university or other training institution with practical experience in a functioning SA/OR group. One illustration of the success of this method is the NATO (North Atlantic Treaty Organization) Fellowships in Operations Research. 14 14 The fellowships, part of a larger training scheme embracing the sciences and engineering, and instituted for the benefit of NATO member countries in the early 19505, were widely publicized in a short brochure explaining what SA/OR is and what it does. The applicant was required to have training in science or engineering and to be proficient in basic mathematics and probability theory. The applicant had to present with his application a letter from his government indicating that on completion of the fellowship there would be a position for the applicant in some part of the government, a position that would use his SA/OR training. A committee of SA/OR experts awarded the fellowships. The candidate presented a list of several acceptable study locations; the awards committee decided where he would go. The committee carefully screened the potential training centers to make sure that the fellow would obtain practical, rather than merely theoretical, training. The fellow was to participate in some purposeful ongoing study. In addition to some six U.S. and U.K. universities (later others in Denmark, France, and Norway), three or four consulting firms with an SA/OR practice were chosen as acceptable training centers. They were

TRAINING SA/OR WORKERS

S9

University-based workshops in the home country may also prove effective. Consideration should be given as well to convening regional conferences.

TRAINING THE SA/OR STAFF The Requisite Skills SA/OR can use a variety of skills, depending on the problem under study. If the group is to work on diverse projects, it will be difficult to maintain a full-time staff with all the necessary skills. Experts from other disciplines should be recruited and fitted into the working structure part-time. They will need little, if any, training in the techniques of SA/OR; what little they do need can be provided by lectures and study groups in connection with the work program. The permanent staff, on the other hand, needs basic skills, including a knowledge of statistics and probability theory, advanced calculus, linear algebra, economics, and some branch of science. These skills are a prerequisite to a formal SA/OR education in most universities. At least one member of the staff must be skilled in the use of a computer; he must be familiar with computer programming and the many techniques of numerical analysis. At least one member should be competent in the area of applied mathematics, although not necessarily trained directly in a formal appliedmathematics course.

Formal Training Formal university work in SA/OR in American universities is usually at the graduate level and leads to either the Master of Science or Doctor of

chosen only after the committee obtained an agreement that the fellow would be given a variety of jobs during his stay, so that he would gain experience in formulation and presentation, as well as in detailed analytic work. The committee rigidly enforced this proviso. The contribution by the participating fmn or university, plus a stipend, provided the fellow a living allowance. The home country was required to contribute minimum travel costs and, in most cases, a portion of the living costs. Fellowships were good for one year, with a possibility of renewal for not more than one additional year. Renewals were granted by the awards committee after a review of the fellow's progress and after receipt of an assurance that his country would still have a useful job waiting for him at the end of the second year. Each fellow and his training center were expected to report on his progress at six-month intervals. Members of a specially appointed panel visited each site to check on conditions, on the progress of the fellows, and on the extent to which the goals of the program were being met.

60

SYSTEMS ANALYSIS AND OPERATIONS RESEARCH

Philosophy degree. Undergraduate degrees may soon be obtainable. Graduate programs usually require work in economics, mathematics, allocation theory, applied stochastic processes, statistics, and computer applications. Various elective areas are included. In addition, students engage in research work on organizational systems and in the design, planning, management, and operation of special systems. Typical applied-research areas include management of research and development; portfolio and fmancial management; site selection; long-range forecasting; queuing systems; production scheduling and inventory control; transport, traffic, and parking systems; water resources; health care and social welfare systems; educational systems; and urban management. Some universities award degrees in SA/OR, under one name or another. Others offer these programs as part of the curriculum in such fields as industrial engineering, systems engineering, electrical engineering, applied mathematics, or computer science.

Short Courses, Seminars, and Workshops

SA/OR training is also provided through seminars, workshops, and short courses. These are usually offered by university faculty and working practitioners through continuing-education programs of universities, research centers, consulting organizations, government bodies, and industrial conferences. Such programs may be broad surveys, may cover elementary or advanced topics, or may explore recent advances in the field. They include discussions of both theory and application.

Participatory Training

The SA/OR fellowship, involving overseas study and work in a functioning SA/OR group, is a good technique for training staff as well as the director. Many staff members have no taste for the entrepreneurial side of SA/OR work, but can benefit, in both theory and practice, from professional contacts with coworkers. Such fellowships may be limited to six months or a year.

6

Technical Assistance Implications

There are three main ways in which the industrialized countries can promote the development of SA/OR in LDCs and help them strengthen their capabilities: . • They can help in the development of LDC university training programs designed to enable the LDCs to train their own people; • They can provide funds to allow their own universities to offer such programs by offering various kinds of training "packages" to LDC students; and • They can provide fmancial and technical help to LDCs in starting their own SA/OR applications, and they can provide moral support for SA/OR professionals in LDCs.

DEVELOPING UNIVERSITY TRAINING PROGRAMS Areas of possible help to LDCs in the development of their own university training programs are curriculum development, exchange professorships or scholarships usable in either the LDCs or the industrialized countries, other kinds of support for study in the industrialized countries, and support for conferences, seminars, workshops, and institutes.

61

62

SYSTEMS ANALYSIS AND OPERATIONS RESEARCH

Curricula Universities in the United States and the United Kingdom have been teaching courses in SA/OR for nearly 25 years; most industrialized countries now offer such courses in one or more of their universities. Professional societies have devoted much attention to the problems of SA/OR education. This experience can be made available to the developing countries, but obviously cannot be transferred without translation with respect to both language and culture. This effort will take time and money, and should be considered for long-term funding.

Exchange Programs A program of exchange professorships should provide a high return on its cost. Exchanges should be planned to span long periods; perhaps the development of "sister universities" is the most effective scheme. The benefits would not be one-sided: students in the industrialized countries would gain a wider understanding of the problems of the developing countries, and their ability to apply SA/OR techniques under a variety of conditions would be enhanced. Since SA/OR relies heavily on the scientific training of its practitioners, programs in science will also contribute to the ability of the developing countries to practice SA/OR. Students in LDCs who are not attracted to laboratory research as such may nevertheless be attracted to a career in science if they become aware that work in SA/OR is possible in their countries; exchange programs can encourage such choices.

Scholarships and Fellowships Most industrialized countries offering the LDCs aid and technical assistance have public and private programs for the award of grants for study at their universities. In the United States, the African Graduate Fellowship Program of American Universities and its Latin American counterpart could serve as channels for the award of SA/OR fellowships to qualified candidates from the two regions. American philanthropic institutions also make fellowship grants either directly or through educational establishments. The Agency for International Development and other bilateral aid agencies, and international and regional aid agencies (United Nations Development Program and United Nations Specialized Agencies, World Bank, and regional development banks, etc.) also give fmancial support for professional study in various specialties. This support is provided either in fulfillment of a basic train-

TECHNICAL ASSISTANCE IMPLICATIONS

63

ing mission or in the context of larger projects with a training component they have funded. There appears to be no lack of channels by which qualified candidates from developing countries can be admitted to institutions, in the United States or elsewhere, to obtain instruction in SA/OR subjects. What may be lacking, on the part of both the existing channels and the potential LOC candidates, is the recognition that SA/OR constitutes a legitimate field of professional specialization and one that can contribute significantly to the development effort of the LOCs.

Conferences, Seminars, and Workshops The aid-giving agencies can assist the development of SA/OR by contributing to the support of a program of conferences, seminars, and workshops. The support can take several forms: financing the publication of minutes and proceedings, paying travel and other costs of foreign experts brought to meetings, or simply providing the occasion for such meetings.

Special Programs A number of U.S. universities have organized special programs in such fields as agriculture and engineering tailored to the particular needs of developing countries. Some of these programs have been undertaken within the framework of a "sister university" relationship. An SA/OR training program could be organized in a similar fashion.

STARTING AN SA/OR PROGRAM Aside from the educational and promotional assistance discussed in the preceding sections, the aid-giving agencies can help an LOC start an SA/OR program through support for equipment purchases, training grants, and advisors, and through provision of SA/OR work as an integral part of feasibility studies sponsored by them.

Equipment, Loans, and Grants It has been said earlier that the equipment requirements tend to be modest. The aid-giving agencies can either make outright gifts of appropriate com-

64

SYSTEMS ANALYSIS AND OPERATIONS RESEARCH

puting equipment or provide soft loans or grants for its purchase. Provision must be made for adequate maintenance, and care must be taken to ensure adequate software· support. The possibility of rental should be kept in mind. Some projects may require observational apparatus such as traffic-counting devices or aerial-survey equipment.

Technical Advisors Foreign technical help may be valuable in the early stages of SA/OR. The aid-giving agencies can pay the salaries and living expenses of experts who will act as working participants and advisors to a beginning SA/OR group. The term of engagement should be long enough for the experts to transfer their skills effectively. This program should be integrated with other programs aimed at the development of planning skills.

LDC Applications for Project Loans and Grants Although it may not be feasible to require that all project loan and grant applications be prepared with the assistance of an SA/OR group, it is reasonable to suggest that they take account of the factors that an SA/OR group would consider, and that the effect of these factors be referred to in the application. Such a suggestion will go a long way toward stimulating the development of local SA/OR capabilities.

Glossary

Some technical terms in this report may be unfamiliar to some readers. The following brief explanations may be useful even though the defmitions may not be complete enough to satisfy a professional SA/OR worker. Algorithm. A procedure for solving a problem; a prescription for a series of arithmetical calculations to be performed one after another, leading from the original data to the solution. Allocation Theory. Deals with the problem of allocating scarce resources so that some preconceived goal is approached as closely as possible. Several different methods are appropriate to different circumstances: e.g., linear programming (q.v.), mixed-integer programming, and dynamic programming (q.v.). No one method offers a solution to all allocation problems. Applied Systems Analysis (ASA). ASA is not merely a technique or group of techniques such as probability theory or mathematical programming; rather it can be thought of as a broad research strategy-a strategy that involves the use of techniques, concepts, and a scientific, systematic approach to the solution of complex problems. It is a framework of thought designed to help decision makers choose a desirable (or in some cases a "best") course of action. The approach may entail any of the following steps: 65

66

GLOSSARY

• Recognizing the existence of a problem or of a constellation of interconnected problems worthy of, and amenable to, analysis. • Denning and bounding the extent of the problem area. It is necessary on the one hand to simplify the problems to the point of analytic tractability and on the other to preserve all vital aspects affected by various possible solutions. The difficult judgment concerning the inclusion or exclusion of problem elements-balancing their relevance to the analytic grasp of the situation against their contributions to unmanageable complication-often determines the success of systems research. • Identifying a hierachy of goals and objectives and examining value trade-offs. • Creatively generating appropriate altematives for examination. • Modeling the complex, dynamic interrelationships among various facets of the problem, taking cognizance of the inherent uncertainties, and realizing that decision rules incorporating feedback mechanisms can be employed. The complexities of many problems can be managed only through the use of modem, high-speed computers. Other problems of a lower level of complexity can be more efficiently handled by automated means than through human control. ASA thus pays heed to many aspects of man-machine interactions. • Evaluating the potential courses of action and investigating the sensitivity of the results to the assumptions made and to facets of the problem excluded from the formal analysis. • Implementing the results of the analysis. Precisely because ASA is a rational approach rather than a technique, the list of steps above should be understood in a qualined sense. Not all the steps need be included in every instance of responsible systems analysis. Some steps may be handled in a more formal manner than others. The order of the steps can be varied or iterated in various patterns-thus the importance of excluded factors may be repeatedly reassessed or the relevance of the objective structure periodically examined. The most useful model will mimic reality with sufficient precision to serve a broad spectrum of decisions and decision makers. The decision stage may therefore be diffuse and broad and follow the completion of the formal scientifIc analysis. In summary, ASA aspires to promote good decision making. As a process, it is intended to force hard thinking about large, complex problems. Where feasible, it makes use of automated techniques for computation and decision resolution. Dynamic Programming. The mathematical theory of multi-stage decision process; the application of this theory. Dynamic programming leads

GLOSSARY

67

by steps to an optimal policy, in ways that often correspond to methods by which decisions are made. The method has wide technical applications and often illwninates the nature of the solution in complex problems. There are specific applications in allocation (q.v.) and optimization theory, control theory, game theory, and in all systems characterized by nonlinear objective functions and constraints. Feedback Dynamics. A feedback system (as opposed to an "open" system) is one in which the magnitude of an output variable is compared to some desired magnitude and is then altered by the use of the information derived from the comparison. Almost all naturally occurring systems incorporate some kind of feedback control. Feedback dynamics is a set of techniques for analyzing the causal structures underlying the behavior of feedback systems. Input-output Analysis. The study and use of input-output models. An inputoutput model of an economy, or a segment of an economy, consists of several sectors (e.g., several industries, plus labor, plus other sectors) which may be disaggregated to any desired extent. It is assumed that in producing one unit of output each sector requires a characteristic amount of input from each other sector. For example, 1 ton-mile of railroad transport will require characteristic amounts of fuel, steel, labor, electronic apparatus, etc. Usually, though not necessarily, the characteristic amounts are expressed in terms of money. An input-output table is simply an array of these characteristic amounts. It can be used to determine the consequences of various asswnptions concerning the output of specified sectors and to test whether a proposed plan for all sectors is in balance. Linear Programming. A mathematical technique for fmding a "best" way to accomplish some given purpose. The "best" means yielding the highest (or lowest) value of some nwnerical quantity that is used to measure the extent to which the operation satisfies the goals of the administrator. For instance, one may want maximwn profit, or lowest cost, or maximwn employment, or minimwn use of foreign exchange; the list of possibilities, obviously, is long. It is asswned that various quantities of different elements (material, machine time, money, labor, etc.) may be used (up to some specified upper limit determined by the capacity of the ~ystem), and that each unit of each element has associated with it a "cost" related to the factor that the administrator wishes to control. The linear programming method is a systematic procedure for trying various combinations of elements in such a way that the control variable comes nearer to its goal (maximwn or minimwn as the case may be) at each try, always keeping within the established limit. The

68

SYSTEMS ANALYSIS AND OPERATIONS RESEARCH

word "linear" is used to mean that the input requirements are directly proportional to the output magnitude, and that the cost of an input is directly proportional to the amount used. When the quantities involved are limited to integer values, the problem is said to be one of "integer programming." Macroeconomic. This term is used in opposition to its conjugate, microeconomic (q.v.). Macroeconomics is the economic study of the structure and behavior of large systems. It may deal with such concepts and magnitudes as gross national product, national income, consumption, investment, and trade balances. Microeconomic. Microeconomics deals with the composition of small economic systems. It attempts to explain how resources are allocated in the production of particular goods and services, and how those goods and services are distributed among the population. ModeL A picturable collection of concepts of some system or operation, together with the defmitions and statements of interrelationships among the concepts. Any formal conceptual picture of the world, or of a segment of it, is a model. In a scientific model the concepts are associated with quantitative measurements and the interrelationships are stated as mathematical equations. In the SA/OR field, "model" is used in several loosely related senses. It is best understood by considering a number of examples, such as those given in Chapter 2. In general, a model is a simplified description of a real system or operation in which features not essential to the study of the problem at hand are omitted. Queuing Theory, or Waiting Line Theory. Deals with systems in which the essential features are 1) a service facility or server, and 2) a queue, or waiting line, of "customers" awaiting service. The mathematical theory relates the length of the waiting lines to the rate of arrival of new customers and the rate at which the server disposes of customers who are being served. Most queuing models explicitly account for randomness in the arrival of customers and randomness in the service time. Queuing theory has a wide range of application to such diverse phenomena as seaports and incoming ships, telephone exchanges and telephone calls, health services and patients, mechanics and machines awaiting repairs, and airstrips and incoming airplanes. Many problems in queuing theory are treated by simulation (q.v.). Random Number. A random number is a number that results from a nonordered process such as throwing dice, or spinning a roulette wheel. Random numbers are a part of the mathematical apparatus of applied SA/OR, used primarily in simulation processes, where they represent the effects of chance causes.

GLOSSARY

69

Pseudo-random nwnbers, created by arithmetical procedures, are a substitute for naturally occurring random nwnbers. In particular, they are generated by electronic computers for use in simulation programs. Simulation. A method for imitating the reaction of a process, or a system, by mathematical computations. For stochastic processes (q.v.), random nwnbers (q.v.) are needed. The mathematical laws that govern the behavior of the system (or model) also prescribe the way in which the state of the system at anyone time depends on the state at a preceding instant. The simulation process is illustrated by the techniques used to deal with the bus-scheduling program of example I, Chapter 2. In a simplified version, one supposes that all buses run exactly on time and the only variable element is the arrival rate of passengers. For each bus stop, and for each hour of the day, an average arrival rate is determined by observation. Then a scheme for using random numbers is devised so that the drawing of a random nwnber at a given bus stop imitates the chance arrival of customers at that stop. At the end of the simulation, the simulator will have a record of the nwnber of passengers on each bus and the nwnber kept waiting. He can therefore calculate the average bus occupancy and the average nwnber of passengers denied boarding. With an electronic computer a day's run can be calculated in a few seconds; the simulation of an entire bus system can be done quickly. It is thus possible to test a schedule for its effect on the riders and on the efficiency of operation of the bus fleet. Site Selection. The process of deciding where to locate an industrial site, factory, seaport, etc. SA/OR can assist the decision maker by constructing a model of the system consisting of the facility and the factors that interact with it. A simple model is the so-called transportation model, which minimizes the swn of the costs of production and transport of goods by locating the production point at the proper place. Software. A collection of computer programs written and maintained by a computer manufacturer for his machines, or by a computer service agency for the machines it operates. Software is of two kinds: the eX,ecutive, or supervisory, programs that control the computer and enable it to accept programs written by the users; and the utility programs for the solution of problems that recur frequently and can be solved by standard methods. They may be simple or elaborate. Since there are differences, large and small, between the software programs of different manufacturers and service agencies, an SA/OR group must have at least one skilled computer user on its staff. Stochastic Process. A stochastic process is a system considered in its timebehavior aspect. When the future of the system is completely deter-

70

GLOSSARY

mined by its past, the system is said to represent a deterministic process. When the future qepends on the past only in a probabilistic way, the system is said to represent a stochastic process. If the effects of chance are small in relation to the effects of the dynamic forces in the system, one may treat the system as though it were deterministic while studying its main behavior. When the probabilistic forces are strong, the system should be treated as a stochastic process.

Appendix A

EXCERPT FROM FOREWORD TO WORLD BANK REPORT ON THE INDUS PROJECT 15 The assignment by the President of the World Bank to undertake a Study of the Water and Power Resources of West Pakistan, with a view to clarifying their potential for development and identifying investment priorities, was a difficult but most rewarding experience for me and my principal assistants. Our work began with a review of existing reports and studies; we concluded that these provided an inadequate basis for judgment in regard to the many issues involved. We further recognized the limitations of what might be called the "traditional" approach to investment planning-one that proceeds via the summation of various economically viable projects, individually evaluated. Even a cursory look at the problems of the Indus Basin indicated that a high degree of interdependence exists between different development and investment decisions. There is, in brief, competition for scarce resources: whether they be, for example, water, capital, general fanning inputs, or entrepreneurial skills. The Study, therefore, would have to be comprehensive in every sense of the word. It would entail not only a careful examination of all aspects of Up. Lieftnick, A. Sandove, and T. Creyke, Water and Power Resources otWest Pakistan-A Study in Sector Planning. World Bank Study Group. 3 vols. (Baltimore: John Hopkins Press, 1968)

71

72

APPENDIX A

agriculture but also a real understanding of the whole irrigation system in all its intricate complexity, including the integration of programs for the simultaneous exploitation of both groundwater and surface water. It would also involve a detailed investigation of West Pakistan's energy needs and resources over the long term with a view to drafting the broad design of an integrated system of generating plants and transmission networks for meeting those needs. Moreover, it was necessary to consider the available resources against the background of the overall economic objectives of Pakistan in order to assist in formulating an efficient strategy of development. We soon realized that the magnitude of the Study made it well suited to the use of modern programming techniques-made possible by the use of computers-by which many projects can be viewed as components of a larger sector for which only restricted resources are available. As our conceptions crystallized, we came to regard investment proposals not only in terms of their general priority, but in terms of their specific strategic place in the development of a complex system of economic interactions. And the broader the scope of the program evaluations, the greater became the requirements for data, as compared to what could be considered adequate for conventional project evaluations, and the more crucial became the selection of specific analytical methods. Equally, the general procedures adopted made for a long and rather discursive report. This published document could have merely presented the main conclusion and recommendations, with supporting arguments. However, because we felt that the methodological and analytical approach would be of wide interest, it was decided to publish large parts of the report in their entirety. This Study has been made possible only by intensive international cooperation of men of wide international experience and experts in modem technology and analysis. Close contact was maintained throughout the course of the Study between the Group, which formed my immediate team, our consultants, the Pakistani authorities, and their consultants. We have had many rounds of discussions, both formal and informal, in Pakistan and the United States and at the European headquarters of the consultant firms involved. These frequent contacts and the fruitfulness of them were practical expression of the spirit of international cooperation which was at the' heart of the Study. I would like here to express my deep appreciation for the assistance afforded us by the Pakastani authorities and by our and their consultants. Our consultants accomplished a major task in preparation of their comprehensive reports; totaling more than 30 volumes, and their knowledge and technical judgment were invaluable to the Bank Study Group. PETER LIEFTINCK

Appendix B

BIBLIOGRAPHY FOR TABLE 1 I. Industry

1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11.

Beattie, D. W. 1973. Improving the structure of a distribution system. Operational Research Quarterly. 24(3):353-364. Blackett, P. M. S. 1948. The advancement of science. Operational Research. 5(17): 26-28. See also Operational Research Quarterly. 1(1):3-6. Bodnarchuk, A., and P. J. Jeannoit. 1969. A maintenance simulation for complex assemblies. Canadilln Operational Research Society Journal. 7(1): 1-1 5. Cantellow, D. G., R. V. Pitt, R. J. Saw, and J. Hough. 1973. Machine-shop problems: an operational research approach. Operational Research Quarterly. 24(4): 503-525. Demiroache, A. R., A. B. Howell, and T. R. Fowler. 1968. Mathematical models for motor vehicle replacement policies in the federal government. Canadilln Operational Research Society JournaL 6(1): 1-19. Douglas, J. 1970. Replacement for more profits. Construction Methods and Equipment. 52(Mar., Part 1):114-122; (Apr., Part 11):101-112. Fairhurst, J. H. 1972. A case study on the measurement of rate ofretum on investment in joint products: the problem of m products made out of n components on f facilities. Operational Research Quarterly. 23(3) :459-4 76. Greening, J. H., and D. F. Osman. 1970. How do you justify capital equipment? Machinery. 76(Nov.):58-59. Lockett, G. 1973. A model of the hide-leather sequence in a tannery. Operational Research Quarterly. 24(4):491-501. Longbottom, D. A. 1973. The application of decision analysis to a new product planning decision. Operational Research Quarterly. 24(1):9-17. Loomba, N. P. 1973. Inventory control, basic blocks and a taxonomy of selected models. Proceedings ofAmerican Production and Inventory Control Society 14th Annual Internatiofllli TechniCllI Conference, Washington, D.C. pp. 26-43.

73

74

APPENDIX 8

12. Lonnstedt, L. 1973. The use of operational research in twelve companies quoted in the Stockholm stock exchange. Operational Research Quarterly. 24(5):535545. 13. McCulloch, G. A., and R. 8nadyopadhyay. 1972. Application of operational research in production problems in the steel industry. Internati01lll1 Journal of Production Research. 10(1):77-91. 14. Meyer, R. A., Jr. 1971. Equipment replacement under uncertainty. Management Science. 17(11):750-758. 15. Moore, P. G., and H. Thomas. 1973. The rev counter decision. Operational Research Quarterly. 24(3):337-351. 16. Muckstadt, J. A. 1973. A model for a multi-item, multi-echelon, multi-indenture inventory system. Management Science. 20(4):472-481. 17. Tao, C. S., and R. H. Day. 1971. A process analysis model of the U.S. steel industry. Management Science. 17(10): 8588-8608. II. Energy

18. 19. 20. 21. 22.

Anderson, Dennis. 1972. Models for determining least-eost investments in electricity supply. Bell Journal ofEconomics and Management Science. 3(1): 26 7301. 8essiere, F. 1970. The investment 85 model of Electricitli de France. Management Science. 17(4):8192-8211. Masse, P., and R. Gibrat. 1957. Application of linear programming to investments in the electrical power industry. Management Science. 3(2):8149-8158. Petersen, E. R. 1973. A dynamic programming model for the expansion of electric power systems. Management Science. 20(4):656~64. Turgeon, A. 1971. Supervising reservoirs and choosing the most economic size for new hydroelectric installations. Canadian Journal ofOperational Research and Information Processing. 9(3):262-272.

III. Transportation

23. 8ennett, R., C. T. W. Curr, and K. H. Haywood. 1972. The scheduling of trawlers in a fleeting system. InternationaIJournal ofProduction Research. 10(1): 23-33. 24. 8ronzini, M. S., J. H. Herendeen, Jr., J. H. Miller, and N. K. Womer. 1974. A transportation-sensitive model of a regional economy. Transportation Research. 8:45~2.

25. Dobell, A. R., L. D. Taylor, L. Waverman, T.-H. Liu, and M. D. G. Copeland. 1972. Telephone communications in Canada: demand production and investment decisions. Bell Journal of Economics and Management Science. 3(1): 175-219. 26. Ross, D. W., R. C. Sandys, and J. L. Schlaefli. 1971. A computer control scheme for critical intersection control in an urban network. Transportation Science. 5(2):141-160. 27. Rothstein, M. 1971. An airline overbooking model. Transportation Science. 5(2):180-192. 28. Saba, S. L. 1970. An algorithm for bus scheduling problems. Operational Research Quarterly. 21(4):463-474. 29. Steenbrink, P. A. 1974. Transport network optimization in the Dutch integral transportation study. Transportation Research. 8: 11-27. IV. Water Resources 30.

Foster, E. T., Jr., T. C. Chen, J. P. Newton, and E. O. Isu. 1972. Improved river

APPENDIX B

31. 32. 33. 34. 35. 36.

v.

75

basin utilization through systems analysis. Water Resources Bulletin. 8(5):863870. Frankel, R. J. 1974. A systems approach to assessment of rural water supply program effectiveness. Water Resources Research. 10(2):163-170. Gopalakrishnan, C. 1972. Economic growth through water resource development: India. Water Resources Bulletin. 8(3) :459-4 72. Haveman, R. H. 1972. The Economic Performance ofPublic Investments: An Ex Post Evaluation of Water Resource Investments. Johns Hopkins Press, Baltimore. Regev, U., and A. Schwartz. 1973. Optimal path of interregional investment and allocation of water. Water Resources Research. 9(2) :25 1-262. Russell, C. B. 1972. An optimal policy for operating a multipurpose reservoir. Operational Research. 20(6):1181-1189. Takeuchi, K., and D. H. Moreau. 1974. Optimal control of multiunit interbasin water resource systems. Water Resources Research. 10(3):407-415.

Irrigation

37. 38. 39. 40. 41. 42. 43.

Dudley, N. J., and O. R. Burt. 1973. Stochastic reservoir management and systems design for irrigation. Water Resources Research. 9(3):507-522. Glickman, T. S., and S. V. Allison. 1973. Investment planning for irrigation development projects. Socia-Economic Planning Science. 7:113-122. Rose, C. J. 1973. Management science in the developing countries: a comparative approach to irrigation feasibility. Management Science. 20(4):423-438. Smith, D. V. 1973. Decision rules in chance-eonstrained programming: some experimental comparisons. Management Science. 19(6):688-702. Smith, D. V. 1973. Systems analysis and irrigation planning. American Society of Civil Engineers Journal of the Irrigation and Drainage Division. Mar.: 89-107. Steward, I., R. M. Hagan, and W. O. Pruitt. 1974. Functions to predict optimal irrigation programs. A merican Society of Civil Engineers Journal of the Irrigation and Drainage Division. June:179-199. Wu, I-pai, and T. Liang. 1972. Optimal irrigation quantity and frequency. A merican Society of Civil Engineers Journal of the Irrigation and Drainage Division. Mar.:117-133.

VI. Agriculture 44. 45. 46. 47. 48.

Harnden, B. M., P. M. Maher, and G. A. Martin. 1973. Forest fIre detection systems design. Management Science. 20(4): 61 7~ 28. Mann, S. H. 1973. On the optimal size for exploited natural animal populations. Operational Research Quarterly. 21(3):661~71. Meyer, C. F., and R. J. Newett. 1970. Dynamic programming for feedlot optimization. Management Science. 16(6):8410-84 26. Minhas, B. S., K. S. Parikh, and T. N. Srinivasan. 1974. Toward the structure of a production function for wheat yields with dated inputs of irrigation water. Water Resources Research. 10(3):383-393. Ward, R. C., D. A. Link, and W. M. Crosswhite. 1972. An application of network theory to water management in poultry processing. Water Resources Bulletin. 8(3):495-S04.

VII. Health Care Delivery 49.

Esogbue, A. 0.1971. Experiments on scheduling disciplines in surgery: a simulated queueing approach. Opsearch. 8(4):264-280.

76 50. 51. 52. 53. 54. 55. 56.

VIn. 57. 58. 59. 60. 61. 62. 63. 64. 65.

APPENDIX B Ginsberg, A. S., and F. L. Offensend. 1968. An application of clccision theory to a medical diagnosis-treatment problem. Institute of ElectriCllI and Electronic Engineers Tranmctions on Systems Science and Cybemetics. SSC4(3):355-362. Gue, R. L 1966. Mathematical programming models for hospital menu planning. The Journal of Institutional Engineering. XVll(8):395-399. Hall, W. K. 1972. An application of multifunction stochastic service systems in allocating ambulances to an urban area. Operati01lQI Research QuDrterly. 20(3): 558-570. Ledley, R. S. 1969. Practical problems in the use of computers in medical diagnosis. Proceedings of the Institute ofElectrical and Electronic Engineers. 57{ll): 1900-1918. Newell, D. J. 1962. Emergency admissions and the pre-discharge ward. The Hospital. 58:13-15. SondiJc, E. J., and F. L. Offensend. 1971. Toward an integrated methodology for the analysis of health~e systems. Operational Resel/Tch Quarterly. 19(6): 13001322. Volz, R. A. 1971. Optimum ambulance location in semi-rural areas. Transportation Science. 5(2):193-203. Urban Planning and Manaaement Chaiken, J. M., and R. C. Larson. 1972. Methods for allocating urban emergency units: a survey. Management Science. 19(4):Pll o-P13 O. Gardiner, C., and P. W. Ward. 1974. A long range financial resource planning model for a local authority. Operational Research Quarterly. 25{l):55-64. Goldberg, M. A. 1973. Simulation, synthesis and urban public decision-making. Management Science. 20(4) :629-643. Heroux, R. L., and W. A. Wallace. 1973. Linear programming and fmancial analysis of the new community development process. Management Science. 19(8):857-872. Kalba, K. 1974. Urban telecommunications: a new planning context. SocioEconomic Planning Science. 8{l):37-45. Kilbridge, M. D., R. P. O'Block, and P. V. Teplitz. 1969. A conceptual framework for urban planning models. Management Science. 15(6):B246-B266. Kolesar, P., and E. H. Blum. 1973. Square root laws for fue engine response distances. Management Science. 19(12): 1368-1378. Mulvihill, M. E., and J. A. Dracup. 1974. Optimal timing and sizing of a conjunctive urban water supply and waste water system with non-linear programming, Water Resource Research. 10(2):170-175. Neering, M., C. Williams, W. Faught, P. Avioli, J. Maylie, D. Sterling, and R. H. Pantell. 1972. Stanford land use study. Socio-Economic Planning Science. 6(4):409-421.

IX. Education Balachandran, K. R., and D. Gerwin. 1973. Variable-work models for predicting course enrollments. Operational Research Quarterly. 2H3):823-834. 67. McNamara, J. F. 1973. Mathematical programming applications in educational planning. Socio-Economic Planning Science. 7(1): 19-36. 68. Schroeder, R. G. 1973. A survey of management science in university operations. Management Science. 19(8):895-906. 69. Turksen, I. B., and A. G. Holtzman. 1972. Information system design for educational management. Socia-Economic Planning Science. 6(1): 1-20. 66.

APPENDIX B

x.

77

Touriam

70.

Gearing, C. E., W. W. Swart, and T. Var. 1973. Determining the optimal investment policy for the tourism sector of a developing country. Management Science. 20(4):487-497.

XI. Criminal Justice System 71. 72. 73.

Abraham, S. C. 1972. Simulation modelling in criminal justice: an approach and demonstration using system dynamics. Presented at the ORSArrIMS/AIIE Joint National Meeting in Atlantic City, N.J., Nov. 8-10. Blumstein, A., and R. C. Larson. 1969. Models of a total criminal justice system. Operational Research Qwlrterly. 17:199-232. Krendel, E. S., and R. M. Dummer. 1972. Evaluating and planning of a component in the criminal justice system. Socio-Economic Pkmning Science. 6(3):217-226.

Appendix C

SUPPLEMENTAL READING LIST Bowles, S. 1969. Planning Educational Systems for Economic Growth. Harvard University Press, Cambridge, Mass. Cohon, J. L., and D. H. Marks. 1973. Multiobjective screening models and water resource investment. Water Resources Research. 9(4):826-836. Drake, A. W., R. L. Keeney, and P. M. Morse, eds. 1972. Analysis of Public Systems. MIT Press, Cambridge, Mass. Eckhaus, R. S., and K. H. Parikh. 1970. Planning for Growth Multisectoral, Intertemporal Models Applied to India. MIT Press, Cambridge, Mass. Federal Water Pollution Control Administration. 1966. Delaware Estuary Comprehensive Study. U.S. Department of the Interior, Washington, D.C. Gatley, D. 1971. Investment Planning for the Electric Power Industry: A Mixed Integer Programming Approach. Unpublished Ph.D. dissertation, Princeton University. Ghosal, A. 1966. Operations research in the coal industry in India. In Proceedings of the 4th International Conference on Operations Research. John Wiley & Sons, Inc., New York. Harvard University Center for Population Studies. 1972. Land, Water and Agriculture in Bangladesh. Report prepared for United Nations Development Program/International Bank for Reconstruction and Development, Cambridge, Mass. Heady, E. 0., et aL 1973. National and interregional models of water demand, land use, and agricultural policies. Water Resources Research. 9(4):777-791. Holst, W. 1970. Agricultural problems of emerging nations. In The Challenge to Systems Analysis, edited by Grace J. Kelleher. John Wiley & Sons, Inc., New York. Kanai, J. 1967. Mathematical Planning ofStructural Decisions. North-Holland Press, Amsterdam. Kendrick, D. A. 1973. Programming Investment in the Process Industries: A n Approach to Sectoral Planning. MIT Press, Cambridge, Mass. Lawrence, J., ed. 1970. Proceedings of the Fifth International Conference on Operational Research, Venice, 1969. Tavistock Publications Ltd., London.

78

APPENDIXC

79

Lieftinck, P., A. R. Sadove, and T. C. Creyke. 1968. Water and Power Resources of West Pakistan, A Study in Sector pumning. Johns Hopkins Press, Baltimore. Manne, A. S., ed. 1967. Investmentsfor Capacity Expansion: Size, Loeotion, and TimePhosing. George Allen & Unwin Ltd., London. Manne, A., and L. Gomeux. 1973. Multi-level Planning: Case Studies in Mexico. NorthHolland Press, Amsterdam. McKean, R. 1958. Efficiency in Government Through Systems Analysis. John Wiley & Sons, Inc., New York. Meta Systems Inc. 1970. SystemsA1IIllysis of Rural Transportation. Prepared for the IBRD, Cambridge, Mass. Meta Systems Inc. 1971. Systems A1IIllysis in Water Resources Planning. Prepared for the National Water Commission, Cambridge, Mass. Meta Systems Inc. 1973. The Tiber River Basin Study. A Report to the Consiglio Nazionale Delle Richerche, Rome, Italy. Cambridge, Mass. Morse, P. M., and L. W. Bacon, eds. 1967. Operations Research for Public Systems. MIT Press, Cambridge, Mass. Nielsen, R. A., and V. R. Locascio. 1973. Computer assisted planning in the public sector. The Management Advisor. May-June: 34-4 1. Revelle, R., et al. 1964. Report on Land and Water Development in the Indus Plain. The report of the White House-Department of Interior Panel on Waterlogging and Salinity in West Pakistan. White House, Washington, D.C. Rogers, P. 1969. A game theory approach to the problems of international rivers. Water Resources Research. 5:749-760. Rogers, P., and D. V. Smith. 1970. The integrated use of ground and surface water in irrigation project planning. American Journal ofAgricultural Economics. 52:13-24. Rose, C. J. 1973. Management science in the developing countries: a comparative approach to irrigation feasibility. Management Science. 20(4):423-438. Thomas, H. A., Jr., and R. Revelle. 1966. On the efficient use of the high Aswan Dam for hydropower and irrigation. Management Science. 8:B296-B311. Thorbecke, E. 1973. Sector analysis and models of agriculture in developing countries. Food Research Institute Studies in Agricultural Economics, Technology and Development. XII(l):73-89.

Resumen en Espanol

Los gobiernos y los habitantes de los paises en vias de desarrollo estan confrontando un problema de surna importancia en 10 que se refiere ala administracion del cambio profundo y rapido. Existe el deseo urgente de comprimir dentro del tiempo comprendido en una generacion el proceso de industrializacion que los paises desarrollados lograron en diez generaciones. EI cambio paulatino basado en experimentos puede ocurrir yesta ocurriendo sin quien 10 dirija, tal como sucedio en gran parte del Mundo industrializado, pero con grandes desperdicios. Por otra parte, cuando los cambios son rapidos se necesitara la direccion de varias fuentes. Mientras mayor sea el cambio, mayor sera la necesidad de comprension de los factores que 10 promueven 0 se oponen a el, 10 ayudan 0 10 entorpecen. Los cambios que acompatian al progreso tecnico y economico influyen en cada aspecto de la vida y de la cultura: la explotacion de la materia prima y fuentes de energia, el adiestramiento y la organizacion del trabajo, el papel que desemperian las mujeres, el caracter de la vida familiar, la experiencia escolar, el empleo del tiempo libre; la lista es interminable. En una atmosfera cambiante y de incertidurnbre, es menester que los planificadores y los oficiales del gobierno empleen todos los metodos fehacientes para la planificacion, analisis y control. Este informe se refiere a un metodo comprobado para ayudar a los administradores de organismos publicos y privados a hacer sus propias decisiones. Durante los ultimos 35 arios se ha producido un desarrollo de largo al· cance en las tecnicas para aplicar las formas de pensar y trabajar de los

80

RESUMEN EN ESPANOL

81

hombres de ciencia a los problemas que confrontan los individuos encargados de tomar decisiones en el gobiemo, los negocios y otros organismos. Estas tcknicas se conocen bajo distintos nombres, tales como: operaciones de investigacion, anaIisis de sistemas, ciencia de la administracion y cibemchica. Todos estos terminos tienen mas 0 menos el mismo significado y por 10 tanto aqui se combinan bajo el nombre de anlilisis de sisternas/investigacion de operaciones (AS/IO). La anotacion pareceni chapucera pero at parecer no existe una altemativa mejor.• Durante su desenvolvimiento, AS/IO ha cosechado en ciertas ocasiones los resultados de las presunciones exageradas de algunos de sus promotores. Es dificil refutar formalmente tales presunciones, y el grupo no ha tratado de hacerlo, ya que sus miembros opinan que un entendimiento realista de los metodos AS/IO y sus campos de aplicacion sirven para colocar estas presunciones en su perspectiva debida y sera de mas valor para aquellos que escucharan el informe. En pocas palabras se podra decir que AS/IO ayuda al individuo encargado de tomar decisiones a traves de la labor realizada en aquellas areas problema que pueden resultar en medidas cuantitativas y teorias afines. Uno de los principios basicos de AS/IO es que el trabajo debe realizarse en intima cooperacion con las personas que conocen a fondo las particularidades de las costumbres locales, y que la decision final compete a la autoridad conocedora de dichas particularidades y no a un extrano. Sin embargo, los logros de AS/IO son reales, y su crecimiento ha sido significativo. La Federacion Intemacional de Sociedades de Operaciones de Investigacion cuenta con sociedades rniembro en 25 paises tanto desarrollados como en vias de desarrollo. El Instituto de Ciencias Adrninistrativas cuenta con miembros en 77 paises. Mediante un acuerdo entre varios gobiemos se creo recientemente en Austria un Instituto Intemacional para Anlilisis de Sistemas Aplicado.· Con relacion a muchos problemas concretos se ha estructurado una teoda extensa que pondran en practica aquellas agencias y empresas que aUn no 10 han hecho. A nivel inmediato se han estudiado con detenimiento los problemas de control del inventario, programacion de la produccion, adrninistracion del equipo de transporte, operacion del hospital, ubicacion de los centros de servicios a la comunidad (bomberos, policia, iglesias, bibliotecas, bancos, etc.), disposicion urbana, y muchos otros. Se han estudiado asimismo otros problemas de mayor alcance y en mayor escala, y tambien se ha progresado en la asignacion de recursos nacionales (como se vera en los ejemplos citados mas adelante en este informe). Los trabajadores de AS/IO son tecnicos y no determinan la politica; esta es tarea de los oficiales nombrados 0 elegidos-los administradores-quienes *En el glosario se explican los terminos (en ingles) designados por un asterisco.

82

RESUMEN EN ESPANOL

conocen a fondo las caractemticas locales, tanto fisicas como sociales, y a quienes los trabajadores de la AS/IO tienen que reportar. Sus responsabilidades consisten en indicar al administrador las consecuencias de distintas politicas administrativas y buscar, juntamente con eH, la manera de poner en pnictica la politica que se haya escogido. Es natural que mientras mas se sepa acerca de las condiciones ffsicas y sociales de importancia del pais de que se trate, mas utiles y practicos seran sus informes. De ahi la importancia de que la mayor proporcion posible del equipo inicial AS/IO este compuesto por nacionales del pais en cuestion. En este informe se da cuenta de las distintas formas en que se pueden reclutar y adiestrar nacionales y encomendarles la obra que redundanl en beneficio de sus propios paises.

ESCASEZ DE RECURSOS Y AS/IO AS/IO es de particular utilidad cuando existe la necesidad de emplear eficazmente los recursos escasos. Los problemas de asignacion de recursos son mundiales. Estos afectan tanto a los pafses desarrollados como a los que estan en vias de desarrollo. Sin embargo, historicamente los pafses en vias de desarrollo estaban mayormente limitados ala produccion de una sola cosecha 0 materia prima, 0 cuando mucho, a unas pocas. En la actualidad estos quieren explotar tOOos sus recursos. Ellos notan que en un tiempo dado un factor unico (tal como pericia, potencial humano, energia, transporte, material 0 capital) se encuentra cnticamente escaso y que esto impide el progreso. Bajo estas circunstancias los gobiemos de los pafses en vias de desarrollo y planificadores constantemente se encuentran frente a la necesidad de aprovechar al maximo los recursos disponibles en el momento, planear la conservacion de los recursos que se necesitaran en estados mas avanzados de progreso, y balancear las necesidades actuales con las inversiories para el futuro. Este es el terreno donde domina el AnaIisis de Sistemas y Investigacion de Operaciones. AS/IO contribuye a la eficiencia actual y a las mejoras del proceso de planificacion de tres maneras: • Mediante el desarrollo de modelos* y estructuras anaIogas ofrece el entendimiento de los factores pertinentes y su interrelacion. • Mediante caIculos y simulacion* predice las posibles consecuencias de actitudes y politicas altemativas. • Mediante caIculos, experimentos planeados y simulacion contribuye a escoger la mejor forma que conducinl al objetivo. Debe notarse que los valores de AS/IO se basan en la facilitacion del proceso de planificacion y su colaboracion en el control de las operaciones y el desarrollo. Esto no es un sustituto de juicio politico, sino que ayuda al definir los

RESUMEN EN ESPAiiloL

83

Hmites de 10 posible y al predecir los resultados de las acciones 0 bien dar a conocer los Hmites de seguridad de dichas predicciones. Claro esta que hay casos donde el empleo de AB/IO no es apropiado; estos se discutiran en la seccion titulada "Advertencias." En otras secciones de este informe se discuten cinco ejemplos de la aplicacion de AS/IO: el desarrollo del Segundo Plan Quincenal de la RepUblica de Corea; el Estudio del Banco Mundial sobre Recursos de Agua y Energia de Paquistm; un estudio del sistema de cultivo en Mexico; un estudio del horario de los omnibuses en Baroda, India; y un estudio sobre la administracion efectiva de los inventarios de sangre total para hospitales y grupos de hospitales en Estados Unidos. Se pueden encontrar otros ejemplos en la planificacion del desarrollo agricola, diseno de puertos maritimos, administracion de sistemas de transporte, ubicacion de las facilidades de distribucion, y en muchas otras areas. Estas se presentan brevemente en el cuadro 1. (Vease tambien la bibliografia, pig. 73.)

FUNCIONES DE AS/IO EN PAisES EN ViAS DE DESARROLLO No sonmuchos los requisitos que se exigen para que los servicios de AB/IO tengan exito. Un grupo AB/IO viable puede consistir de unos cuatro a seis profesionales, siempre y cuando uno de ellos sea especializado en el uso de com· putadoras y otro en matematica aplicada; los otros miembros podran ser especialistas en ciencias, economia 0 bien matematicos. El grupo debera estar encabezado por un director y debera informar a una persona con autorizacion para tomar decisiones (en este informe tambien se Ie denomina administrador) quien no formara parte del grupo pero se sirve de los resultados obtenidos y colabora con ellos en su trabajo. La efectividad del grupo se determinara por la buena voluntad del individuo autorizado para tomar decisiones para discutir sus problemas libremente y estar disponible para dar asesoramiento. En gran parte el exito depende de la competencia del director del grupo quien debera compartir las preocupaciones del administrador y comprender los metodos de sus colaboradores. Ademas debera poseer la facultad de hacer nuevas preguntas y la habilidad de aplicar las ideas y talentos de su grupo para estar seguro de que estos se usarm en toda su extension. Un grupo AS/IO necesitara los servicios de consultores para complementar sus habilidades tecnicas; por ejemplo, ingenieros civiles cuando se trate del diseno de puertos maritimos 0 redes de carreteras; peritos en econometria cuando se trate de planes nacionales, y expertos en campos tales como agricultura, banca, ecologia y medicina cuando se trate del desarrollo regional. Dichos consultores podran reclutarse localmente 0 en el extranjero segun sea la necesidad para proyectos especificos. El grupo debera tener acceso, cuando sea necesario, a una computadora

84

RESUMEN EN ESPANOL

grande y nipida. (La computadora pOOra ubicarse en cualquier parte, hasta en un pais extranjero.) Si se les da la seguridad que tendran dicho acceso, el grupo no tendra que poseer una maquina grande. Mucho del trabajo experimental y mucha de la preparacion de datos puede realizarse en maquinas pequeJias con una inversion modesta de capital.

ADIESTRAMIENTO PARA PARTICIPAR EN UN PROGRAMA AS/IO En la mayoria de los programas no graduados en Estados Unidos se requiere de los estudiantes del programa AS/IO que tengan conocimiento sobre estadistica y teona de probabilidad, caIculo avanzado, algebra lineal, economia y alguna rama de las ciencias naturales. Tambien es deseable que hayan tenido cursos sobre programacion de computadora y ciencias sociales. Los programas a nivel graduado por 10 general requieren matematica, teoria de asignacion,· procedimientos de estocastica aplicada,· administracion de investigacion y desarrollo, administracion financiamiento, seleccion dellugar,· proyecciones a largo plazo, sistema de espera y teona de inventario, transporte, sistemas de trafico y estacionamiento, administracion de recursos hidraulicos, sistemas de servicios de salud y prestacion de servicios de beneficencia, y sistemas educativos y urbanos. Si bien algunas universidades de paises en vias de desarrollo que siguen aun bajo la influencia europea se han mantenido separadas de trabajo practico, se reconoce con mayor fecuencia que tanto los miembros de la facultad como los estudiantes deben ser desafiados por los problemas de la vida. El desarrollo de las aptitudes de AS/IO pueden contribuir a promover esta tendencia de la manera siguiente: • Por medio de conferencias locales y regionales. El personal docente de las universidades puede ser expuesto a las oportunidades profesionales que existen en AS/IO asi como a las necesidades de sus paises. • Las AS/IO pueden dirigirse a los miembros de la facultad para consultarlos asi como para discutir con ellos el desarrollo de cursos de adiestramiento en los cuales se incluyen problemas practicos como parte del programa. • Un programa de becas para estudio 0 trabajo con grupos AS/IO en paises industrializados puede despertar una consciencia del papel practico que puede desempeJiar un miembro de la facultad con orientacion cientffica. Tanto el prestigio como la rivalidad que estan asociados con el programa de becas pueden contribuir a promover el conocimiento de la naturaleza y valor de AS/IO.

85

RESUMEN EN ESPANOL

OBSTACULOS AL DESARROLLO Y

usa DE AS/IO

Posiblemente el obstaculo principal que se encuentra en el uso de AS/IO es la falta de conocimiento por parte del adrninistrador en cuanto a su naturaleza y aplicaciones, 10 cual se expresa tipicamente en las declaraciones que van desde poco interes manifestado hasta la actitud totalmente negativa como los informes Sagasti que se han recibido de oficiales del gobiemo 1 : Las instituciones gubemamentales son complejas y la burocracia es sumamente despaciosa. Se encuentran involucrados muchos problemas politicos los cuales no se pueden resolver mediante el uso de AS/IO. AS/IO podni ser util pero no en la actualidad, puesto que es muy caro y los datos necesarios para aplicarlo no estan disponibles. Sagasti encontrolas siguientes declaraciones que representan la actitud de los ejecutivos industriales: Los linicos problemas urgentes son puramente tcknicos. El riesgo del fracaso es alto y se necesita la garantia de que los resultados seran positivos. Se ve claramente que AS/IO no podra tener exito si no cuenta con asesoramiento especifico que tenga relacion directa con su aplicaci6n. Existe el peligro de no seleccionar debidamente las solicitudes iniciales 10 cual redundara en resultados desalentadores. Parte de la dificultad estriba en que con frecuencia el procedimiento total AS/IO se confunde con tecnicas especificas que se emplean en el mismo. El usuario potencial muy a menudo da por sentado que AS/IO significa programacion linear· (0 simulacion, 0 teona de espera, 0 cualquier otra cosa), y comprende que su problema no se puede resolver empleando esa tecnica en particular. Las actitudes conservadoras y la resistencia que emanan de la poca 0 ninguna familiaridad pueden sobreponerse solamente con el correr del tiempo y el desarrollo de ejemplos visibles, fehacientes y locales de trabajo positivo. Por 10 tanto, en la mayona de los paises en vias de desarrollo cabe la posibilidad de que AS/IO se desarrolle muy despacio al principio y que luego se desarrolle con mayor velocidad a traves de un periodo de lOa 20 anos. Esto es en verdad 10 que sucedio en los paises industrializados. Otro obstaculo podra ser la falta de operadores debidamente adiestrados. Esto es de menor consecuencia; los trabajadores AS/IO pueden formarse en un tiempo relativamente corto. Se podra contratar a personal extranjero 1 Francisco R. Sagasti. Management Sciences in an Underdeveloped Country: The Case of Operations Research in Peru. (Philadelphia: Management Science Center, University of Pennsylvania, 1971).

86

RESUMEN EN ESPANOL

especializado para dirigir durante las primeras fases de la puesta en pnictica del programa mientras que un grupo local se prepara para llevar la responsabilidad total. AI escoger a tales especialistas, se debe tener sumo cuidado de que ellos tienen la experiencia en AS/IO y que pueden transmitirla a los aprendices. Una vez que se comprenda debidamente el concepto AS/IO parece ser que la falta de informacion no es un obstaculo que no se pueda superar. La primera fase de la labor de investigacion definira cuat es la informacion necesaria y revelara como se obtendra. Donde no exista la informacion, aun las observaciones mas rudimentarias con frecuencia produciran resultados de valor. Es decir, mejor es una cantidad minima de informacion que no contar con ninguna.

POSIBLES BENEFICIOS Con frecuencia se ha discutido, 0 temido, que los recursos de habilidades y dinero dedicado a AS/IO competiran con otros usos esenciales de los rnismos talentos y fond os y que el costo del empleo de AS/IO sera elevado. Si se considera el empleo de AS/IO meramente como la prosecucion de estudios aislados en los cuales no participan los administradores 0 personal ejecutivo y cuyos resultados no son mas que informes para decorar las bibliotecas, se ha justificado el punto de vista sombrio. Por el contrario, si AS/IO se realiza como un esfuerzo mancomunado por parte de hombres de ciencia, investigadores con una orientacion cientifica y personal ejecutivo de accion, el ahorro de los recursos justificara en gran medida la inversion. La aplicacion de las tecnicas AS/IO redundara en mejores metodos de ingenieria, fondos para desarrollo, inversiones de capital y esfuerzos humanos.

RESULTADOS De las investigaciones realizadas por el grupo sobre las aplicaciones de AS/IO en paises en vias de desarrollo se desprenden las conclusiones siguientes: • AS/IO es una tecnica comprobada para asistir a los administradores y a otras personas con poder de decision para resolver problemas apropiados. • Es de valor particular cuando los recursos limitados deben usarse para que rindan el maximo provecho. • Su aplicacion esta al alcance de la mayoria de los paises en vias de desarrollo.

RESUMEN EN ESPANOL

87

• Hombres de ciencia locales y tecnicos pueden desarrollarlos y ponerlos en practica. • Los paises industrializados pueden ayudar a los que estan en vias de desarrollo mediante el adiestramiento de personal AS/IO y el desarrollo de la operacion AS/IO. • AS/IO no es una panacea para los problemas de los paises en vias de desarrollo; se deberan tomar en cuenta sus limitaciones (vease "Algunas advertencias," pig. 53).

RECOMENDACIONES El grupo recomendo que los paises en vias de desarrollo presten atencion a 10 siguiente: • Establecer un grupo prototipo AS/IO que informara a un administrador encargado de que este tenga exito. • Escoger con cuidado a los especialistas que serviran de asesores para 10 coal emplearan pruebas de competencia y el consejo de practicantes AS/IO de experiencia. • Promover la inclusion de la practica de AS/IO y asignaturas afines en cursos universitarios. • Patrocinar reuniones para que el personal ejecutivo de la industria privada y del gobiemo se puedan familiarizar con AS/IO. • Promover la adjudicacion de becas a posibles practicantes de AS/IO para estudiar en el propio pais 0 en el extranjero. El grupo hizo las recomendaciones siguientes a aquellas agencias que esten ofreciendo ayuda: • Alentar la formulacion de planes de proyectos integrados que se basen sobre el estudio AS/IO. • Facilitar fondos para becas y participacion en AS/IO. • Contribuir a la promocion de conferencias nacionales y regionales relacionadas con la teoria y aplicacion de AS/IO. • Apoyar a las sociedades AS/IO locales y regionales. • Apoyar directamente a ciertos proyectos y programas AS/IO.

Resume en Francais

La fa~on dont les gouvernements et les peuples des pays en developpement peuvent faire face Ii un changement profond et rapide est un probleme capital auquel lis se trouvent confrontes. Us manifestent l'ardent desir que Ie processus d'industrialisation tienne dans l'espace d'une generation, quand il en a fallu dix aux nations les plus avancees. Un lent changement, fonde sur des Catonnements, se poursuit sans directives, comme ce fut Ie cas dans maint pays industrialises-mais il mene au gaspillage. Au contraire, s'il est rapide il peut necessiter d'etre pris en main de differents cotes. Plus il est vaste, plus il faut comprendre les facteurs qui Ie stirnulent ou lui font opposition, qui l'aident ou lui font prendre une mauvaise voie. L'evolution qui accompagne Ie progres technique et economique touche tous les aspects de l'existenceet de la culture: exploitation des matieres premieres et des ressources energetiques, formation et organisation des travailleurs, role de la femme, nature de 13 vie de famille, sco13rite, occupation des loisirs ... la liste est interminable. Dans un climat de changement et d'incertitude, les planificateurs et les hauts fonctionnaires doivent se servir de toutes les methodes fiables pour organiser, analyser et diriger. Ce rapport traite d'une methode qui a fait ses preuves et qui permet aux dirigeants des organismes prives et publics de prendre leurs decisions en connaissance de cause. Au caurs des trente-cinq dernieres annees, on a assiste Ii de grands progres dans l'application des methodes de pensee et de travail communement employees par les scientifiques pour resoudre les problemes auxquels avaient Ii

88

REsUME EN FRANCAIS

89

faire face les dckideurs des gouvernements, du monde des affaires et de differents organismes. Suivant Ie cas, ces methodes sont connues sous Ie nom de : recherche operationnelle, analyse des systemes, science du management et cybernetique. Tous ces terrnes signifient a peu pres la meme chose et its se trouvent combines ici sous la designation generale de : "analyse des systemesrecherche operationnelle", qui apparaitra desorrnais dans ce rapport sous Ie sigle "AS-RO" qui, il faut Ie dire, est mal commode mais on n'a rien trouve demieux.* A certains moments de son developpement, quelques tenants de l'AS-RO ont pretendu qu'elle etait la panacee. On pourrait refuter cette assertion mais Ie groupe d'experts ne s'y est pas hasarde, car ses membres sont d'avis qu'il est plus avantageux, pour lui donner ses vraies dimensions, que Ie public auquel s'adresse ce rapport comprenne de falt0n realiste ses methodes et ses domaines d'application. En un mot, on peut dire que l'AS-RO apporte son aide au decideur dans les domaines qui peuvent fournir des mesures quantitatives et des theories apparentees de la plus grande utilite. Un de ses principes fondamentaux est que Ie travail doit 'etre realise en collaboration etroite avec ceux qui sont au courant des caracteristiques locales et que la decision finale doit etre prise non pas par un profane mais par un responsable qui connmt ces particularites. Toutefois, la reussite de l'AS-RO est reelle et son developpement important. La Federation internationale des societes de recherche operationnelle (IFORS) compte des societes-membres dans plus de vingt-cinq pays industrialises ou en developpement. L'Institut des sciences du management compte des membres individuels dans soixante-dix-sept nations. Un Institut international pour l'application de l'analyse des systemes* a ete recemment cree en Autriche a la suite d'un accord entre plusieurs gouvernements. On a compile un ensemble theorique important, portant sur de nombreuses questions concretes, qui est pret a etre utilise par les organismes et les entreprises qui n'en ont pas encore fait usage. Dans l'immediat, on a minutieusement etudie de nombreuses questions dont celles ayant trait, en particulier, a la verification des stocks, aux previsions de la production, a l'exploitation des transports, au fonctionnement des hopitaux, a l'implantation des institutions locales (pompiers, police, eglises, bibliotheques, banques, etc...) et aux plans d'urbanisme. On s'est aussi attaque a des problemes a long terrne, de plus grande envergure, et des progres ont ete realises en matiere de repartition des ressources nationales (comme on pourra s'en rendre compte ala lumiere des exemples figurant plus loin dans ce rapport). Les chercheurs se consacrant a l'AS-RO sont des techniciens. lIs n'ont pas ·On trouvera dans Ie glossaire I'explication des termes (en anglais) techniques marques d'un asterisque.

90

RESUME EN FRANc;AIS

a prendre de decisions qui sont du ressort des fonctionnaires elus ou nommes, c'est-a-dire des administrateurs connaissant les caracteristiques locales, tant materielles que sociales, et sous la supervision de qui ils sont places. Leur tache consiste a indiquer a l'administrateur les consequences des differentes politiques envisagees et, de concert avec lui, a determiner les moyens par lesquels la politique choisie pourra etre mise en oeuvre. Plus leur connaissance des particularites materielles et sociales propres au pays interesse sera grande, plus leurs rapports seront utiles dans leurs aspects pratiqueso Pour cette raison, il est essentiel que Ie plus grand nombre possible de chercheurs de l'AS-RO soient des ressortissants du pays considereo Ce rapport indique un certain nombre de moyens par lesquels ces ressortissants peuvent etre recrutes, formes et mis au travail au plus grand benefice de la nation.

RESSOURCES PEU ABONDANTES ET AS·RO L'AS-RO est d'une utilite toute particuliere quand on doit tirer Ie meilleur profit de ressources limitees. L'allocation optimale des ressources est un probIeme mondial. 11 a, en effet, une resonance aussi bien dans les pays industrialises que dans ceux en developpement. Jadis, toutefois, les pays nonindustrialises se contentaient d'exploiter un seul type de culture ou de matiere premiere, ou meme de quelques-uns, en mettant les choses au mieux. Aujourd'hui, ils veulent utiliser la totalite de leurs ressources. IIs s'aper~oivent qu'une composante : competence, main-d'oeuvre, energie, transports, matieres premieres ou capital, fait preuve, a un moment donne, d'une penurie grave entravant ainsi Ie progreso Dans ces conditions, les gouvernements et les planificateurs des pays en developpement doivent constamment faire face a la necessite d'employer au mieux les ressources dont ils disposent, de prevoilla conservation de celles dont ils auront besoin pour les etapes futures du progres et de faire la part des besoins actuels par rapport Ii ceux de l'avenir. C'est la Ie domaine d'application de l'analyse des systemes et de la recherche operationnelle qui contribue actuellement de trois fa~ons differentes a l'efficacite et a l'amelioration du processus de la planification: • Grice a la mise au point de modeles· et d'analogues elle fournit une comprehension des parametres applicables et de leur action reciproqueo • Grace au calcul et a la simulation· elle prevoit les consequences possibles des politiques et des mesures eventuelles. • Grace au calcul, a une experimentation planifiee et a la simulation elle permet de choisir Ie meilleur moyen d'atteindre l'objectif.

RESUME EN FRA~AIS

91

II est essentiel de noter que sa valeur reside dans Ie fait qu'elle facilite Ie processus de planification et qu'elle aide Ii superviser les operations et Ie developpement. Elle ne peut pas se substituer au jugement politique, mais elle est precieuse pour definir les frontilhes du possible et prevoir Ie resultat des actions entreprises ou, tout au moins, pour etablir la limite de certitude de ces previsions. Bien entendu, il y a des cas ou elle est employee Ii mauvais escient; on en parlera dans la section "Mises en garde". Plus loin dans ce rapport on examine cinq exemples d'application de l'AS-RO: Ie developpement du deuxieme plan quinquennal de la Coree du Sud, l'etude de la Banque mondiale sur les ressources en eau et en energie electrique du Pakistan, une etude des systemes de culture vivriere au Mexique, une etude sur les horaires des autobus de la ville de Baroda en lode et une etude americaine sur la gestion efficace de stocks de sang entier pour les hopitaux ou groupes d'hopitaux. On peut trouver d'autres exemples dans la planification du developpement agricole, la conception de ports maritimes, l'exploitation de reseaux de transport, l'implantation d'installations de distribution (depots, stations de groupage), les previsions de la production et, aussi, dans de nombreux autres domaines. Ces exemples sont sommairement indiques au tableau I, page 34. (Voir aussi la bibliographie, page 73.)

PRATIQUE DE L'AS·RO DANS LES PAYS EN DEVELOPPEMENT Une pratique couronnee de succes de I'AS-RO ne represente pas d'exigences excessives. Un groupe efficace consacre Ii cette discipline peut comprendre seulement quelques personnes qualifiees pour autant que, parmi elles, l'une ait l'habitude d'utiliser un ordinateur electronique et qu'une autre connaisse les mathematiques appliquees; Ie reste pouvant etre constitue de scientifiques, d'economistes ou de mathematiciens. Le groupe doit avoir Ii sa tete un chef qui est place lui-meme sous la supervision d'un decideur-aussi appele adrninistrateur dans ce rapport-qui ne fait pas partie du groupe mais met Ii profit ses decisions et collabore avec lui. L'efficacite de ce groupe sera fonction de la bonne volonte qu'apportera Ie decideur Ii discuter librement de ses probIemes et Ii donner des conseils. La reussite depend aussi largement de la competence du chef de groupe qui doit partager les soucis de l'administrateur et comprendre les methodes de travail de ses collegues. II doit aussi faire preuve d'imagination et poser de nouvelles questions et avoir la capacite de tirer Ie maximum des idees et des talents de son groupe. Un tel groupe aura besoin d'utiliser des consultants qui completeront ses propres capacites techniques: des ingenieurs T.P. quand il s'agira de concevoir

RESUME EN FRAN~AIS

92

des ports maritimes et des reseaux routiers, des specialistes de I'econometrie quand il s'agira de plans Ii l'echelle nationale et d'experts dans Ie domaine de I'agriculture, de la banque, de l'ecologie et de la medecine quand il s'agira de developpement regional. Ces consultants peuvent etre recrutes sur place ou Ii l'etranger, suivant les besoins de projets particuliers. I.e groupe doit avoir acces, chaque fois qu 'ille faudra, un ordinateur de grande dimension et rapide qui peut se trouver n'importe ou, meme dans un pays etranger. Ainsi, Ie groupe n'aura pas besoin de posseder sa propre machine de grande taille. La majeure partie du travail experimental et de la preparation des donnees peut etre executee sur de petites machines ne representant qu'un faible investissement fmancier .

a

.

FORMATION EN MATIERE D'AS-RO

a

La plupart des programmes universitaires americains I'intention des etudiants d'AS-RO exigent de ces derniers qu'its aient une connaissance de la statistique et de la theorie des probabilites, du calcul avance, de l'algebre lineaire, de I'economie et d'une branche quelconque des sciences naturelles. Des cours de programmation d'ordinateur et de sciences sociales sont egalement souhaitables. D'habitude, les programmes de cours universitaires avances comprennent les mathematiques, la theorie de I'allocation optimale des ressources,· les processus stochastiques appliques,· la gestion de la recherche et developpement, la gestion financiche et de portefeuilles, Ie choix des sites,· la prevision Ii longue echeance, la theorie de la me d'attente· et des stocks, les transports, la circulation et Ie stationnernent, la gestion des ressources hydrauliques, les systemes assurant les services rnooicaux et sociaux et les systemes educatifs et urbains. Alors que, dans les pays en developpement, certaines universites sous I'influence europeenne se sont traditionnellernent cantonnees dans la theorie, on reconnal't de plus en plus que les professeurs et les etudiants ont besoin de se trouver confrontes Ii la pratique. La creation de competences dans Ie domaine de l'AS-RO peut aider de differentes fa~ons Ii prornouvoir cette tendance:

• Grace Ii des reunions au niveau local et regional, Ie corps enseignant des universites peut etre rnis au courant des possibilites qui s'offrent dans l'AS/ RO et se trouver confronte aux besoins du pays. • I.es groupes AS-RO peuvent s'adresser Ii des membres du corps enseignant pour consultation et pour rnettre sur pied des cours de formation comprenant des questions pratiques. • Des programmes de bourses d'etudes et de travail dans des groupes d'AS-RO des pays industrialises peuvent donner conscience du rOle pratique

RESUME EN FRAN~AIS

93

a

que peuvent jouer les membres du corps enseignant vocation scientifique. Le prestige, joint la notion de concours, qui s'attache un programme de bourses pourrait promouvoir la prise de conscience de la nature et de la valeur de l'AS-RO.

a

EMPECHEMENTS DE L'AS-RO

a

ALA MISE AU POINT ET AL'EMPLOI

Le manque de familiarisation de l'administrateur avec la nature et les applications de cette discipline est probablement l'obstacle majeur son em· ploi. Ceci decoule tres clairement de declarations que l'on peut qualifier de tiedes negatives. Sagasti rapporte les declarations suivantes emanant de hauts fonctionnaires: 1

a

a

.

• Les institutions d'Etat sont complexes et la bureaucratie est tres lente. n existe de nombreux problemes politiques qui ne se pretent pas la solution par l'AS-RO. • L'AS-RO pourrait etre utile, mais pas maintenant. C'est une discipline trop couteuse et les donnees indispensables son application font defaut.

a

a

n a decouvert, egalement, les declarations suivantes qui sont representatives de l'attitude des chefs d'entreprises et des industriels: • Les seuIs problemes urgents sont d'ordre purement technique. • Les risques d'echec sont eleves et i1 faut que Ie succes soit garanti. • nest naturellement impossible que l'AS-RO reussisse sans que des capacites particulieres soient liees son application. On peut craindre que les applications initiales soient mal choisies et que les resultats en soient donc decourageants.

a

Une partie de la difficulte reside dans Ie fait que Ie processus complet de l'AS-RO est souvent confondu avec les techniques specifiques dont on se sert. L'utilisateur eventuel pense souvent que l'AS-RO signifie programmation lineaire* (ou simulation, theorie de la me d'attente, ou n'importe quoi) et il sait que son probleme ne peut pas etre attaque par cette technique particuliere. Une attitude prudente et une certaine resistance, decoulant d'un manque I Francisco R. Sagasti. Ma1lJlgement Sciences in an Underdeveloped Country: The Case of Operations Research in Peru. (Philadelphia: Management Science Center, University of Pennsylvania, 1971).

94

RESUME EN FRAN~AIS

de familiarisation, ne peuvent etre sunnontees qu'avec Ie temps et grace a des realisations reussies, concretes et fiables sur Ie plan local. Done, dans la plupart des pays en developpement, il est vraisemblable que I'AS-RO se developpera lentement au debut, puis plus rapidement au cours d'une periode s'etendant sur dix a vingt ans. C'est ce qui s'est produit dans les pays industrialises. Un autre obstacle pourrait etre la penurie de professionnels bien entra:1nes. C'est moins grave. Les chercheurs de l'AS-RO peuvent {hre fonnes dans un laps de temps assez court. Durant les premiers stades de realisation du prograqlme, on pourrait avoir recours a des specialistes etrangers alors que dans Ie meme temps les membres du groupe local sont fonnes pour pouvoir prendre la releve. En ce qui concerne Ie choix de ces specialistes, il faut prendre grand soin de s'assurer qu'ils possedent une experience reelle de I'AS-RO et qu'ils sont a meme de la transmettre en guidant les stagiaires. Une fois compris la notion d'AS-RO, il est evident que Ie manque de donnees n'est pas un obstacle insunnontable. La fonnulation du travail de recherche pennettra de definir les,besoins en donnees et indiquera quel effort doit etre fait pour les recueillir. Quand elles manquent reellement, les observations les plus rudimentaires peuvent toutefois fournir des resultats de qualite. En fait, il vaut mieux avoir peu de donnees que pas de donnees du tout.

RENTABILITE EVENTUELLE On a souvent craint que les ressources en personnel qualifie et en fonds destines a l'AS·RO entrent en conflit avec d'autres utilisations primordiales de ces memes talents et de ces memes fonds et que Ie cout d'option de l'AS·RO soit eleve. Si l'on considere simplement son emploi comme la menee a bien d'etudes isolees auxquelles les administrateurs et les cadres superieurs ne participent pas et qui se traduisent par des rapports qui ne servent qu'a decorer les rayons des bibliotheques, une vue pessimiste est parfaitement justifiee. Toutefois, si elle est entreprise comme un effort de collaboration entre des chercheurs a vocation scientifique et des responsables dynamiques, l'economie des ressources justifiera largement l'investissement. L'application de ses techniques se traduira par un meilleur emploi des qualifications techniques, des fonds de developpement, des investissements et des efforts humains.

CONCLUSIONS Les conclusions du groupe d'experts sur les applications de I'AS-RO dans les pays les moins developpes sont les suivantes:

REsUME EN FRAN~AIS

95

• C'est une technique qui a fait ses preuves pour aider les administrateurs et autres dtkideurs n!soudre les problemes devant lesquels ils sont places. • Elle represente une valeur indeniable quand on doit tirer profit au rna· ximum du peu de ressources dont on dispose. • Son application est la portee de la plupart des pays en developpement. • Elle peut etre consue et realisee par les scientifiques et les techniciens locaux. • Les pays industrialises peuvent apporter leur aide aux pays les moins developpes en matiere de formation du personnel et de mise au point de son fonctionnement. • Ce n'est pas la panacee pour resoudre les problemes des pays les moins developpes. Ses limites doivent etre prises en consideration (voir "Quelques mises en garde", page 53).

a

a

RECOMMANDATIONS Le groupe d'experts recommande ce qui suit developpement:

al'attention des pays en

• Mise sur pied d'un groupe prototype- d'AS-RO qui fera son rapport a un administrateur qui est attache son succes. • User d'une grande prudence dans Ie choix de specialistes faisant fonction de conseillers, en mettant en concurrence diverses sources et en prenant l'avis de personnes versees dans l'AS-RO. • Susciter des programmes universitaires sur la pratique de I'AS-RO et de sujets apparentes. • Organiser des conferences destinees familiariser les responsables des affaires et de l'administration avec l'AS-RO. • Encourager l'attribution de bourses, chez eux ou l'etranger, aux praticiens eventuels de I'AS·RO.

a

a

a

Le groupe d'experts recommande ce qui suit d'assistance financiere et technique:

al'attention des organismes

• Appuyer la formulation de projets integres fondes sur l'etude de I'AS-RO. • Foumir des fonds destines aux bourses d'etudes et de recherche en AS-RO. • Aider l'organisation de conferences, au niveau national et regional, sur la theorie et la pratique de I'AS-RO. • Porter assistance aux associations locales et regionales d'AS-RO. • Accorder un soutien direct des programmes choisis d'AS-RO.

a

a

BOARD ON SCIENCE AND TECHNOLOGY FOR INTERNATIONAL DEVELOPMENT

Members

Professor, Department of Entomology and Section of Ecology and Systematics, Cornell UniversitY,lthaca, New York, Chairman, 1976CARL DJERASSI, Department of Chemistry, Stanford University, Stanford, California, Chairman, 1973-1975 GEORGE S. HAMMOND, Foreign Secretary, National Academy of Sciences RUTH ADAMS, American Academy of Arts and Sciences, Boston, Massachusetts JOHN D. BALDESCHWIELER, Division of Chemistry and Chemical Engineering, California Institute of Technology, Pasadena, California IVAN L. BENNETT, JR., Dean, School of Medicine, New York University Medical Center, New York, New York GEORGE BUG LIARELLO, President, Polytechnic Institute of New York, Brooklyn, New York CHAR LES DENNISON, Former Vice President, International Minerals and Chemicals Corporation, New York, New York WILLIAM R. HEWLETT, President and Chief Executive Officer, HewlettPackard Company, Palo Alto, California WILLIAM A. W. KREBS, Vice President, Arthur D. Little, Inc., Cambridge, Massachusetts FRANKLIN A. LONG, Program on Science, Technology, and Society, Cornell UniversitY,lthaca, New York JOSEPH PETT IT, President, Georgia Institute of Technology, Atlanta, Georgia JOSEPH B. PLATT, President, Harvey Mudd College, Claremont, California HELEN RANNEY, Chairman, Department of Medicine, University Hospital, San Diego, California ROBERT M. WALKER, McDonnell Professor of Physics, Director, Laboratory of Space Physics, Washington University, St. Louis, Missouri STERLING WORTMAN, Vice President, Rockefeller Foundation, New York, New York DAVID PIMENTEL,

Staff VICTOR RABINOWITCH,

Director

96

Board of Science and Technology for I nternational Development Commission on International Relations National Academy of Sciences-National Research Council 2101 Constitution Avenue, Washington, D.C. 20418, USA Advisory Studies and Special Reports Published

I.

2.

3.

4.

6.

7.

8.

9.

10.

II.

12.

13.

East Pakistan Land and Water Development as Related to Agriculture. January 1971. 67 p. (out of print) Reviews World Bank proposed action program in land and water management. NTIS Accession No. PB 203-328. $4.25. The International Development Institute. July 1971.57 p. Endorses concept of new science-based technical assistance agency as successor to AID; examines its character, purposes, and functions. Solar Energy in Developing Countries: Penpectives and Prospects. March 1972. 49 p. (out of print) Assesses state of art, identifies promising areas for R&D, and proposes multipurpose regional energy research institute for developing world. NTIS Accession No. PB 208-550. $4.25. Scientific and Technical Information for Developing Countries. April 1972. 80 p. (out of print) Examines problem of developing world's access to scientific and technical information sources, provides rationale for assistance in this field. and suggests programs for strengthening information infrastructure and promoting information transfer. NTIS Accession No. PB 210-107. $4.75. Research Management and Technical Entrepreneurship: A U.s. Role in Improving Skills in Developing Countries. 1973.40 p. Recommends initiation of a systematic program and indicates priority elements. U.s. International Firms and R, D & E in Developing Countries. 1973.92 p. Discusses aims and interests of international finns and developing-country hosts and suggests that differences could be mitigated by sustained efforts by the fmns to strengthen local R, D & E capabilities. Ferrocement: Applications in Developing Countries. 1973. 89 p. Assesses state of the art and cites application of particular interest to developing countries-boatbuilding, construction, food- and water-storage facilities, etc. Mosquito Control: Some Penpectives for Developing Countries. 1973.63 p. Examines biological-control alternatives to conventional pesticides; evaluates state of knOWledge and research potential of several approaches. Food Science in Developing Countries: A Selection of Unsolved Problems. 1974. 81 p. Describes 42 unsolved technical problems with background infonnation, possible approaches to a solution, and infonnation sources. Aquatic Weed Management: Some Perspectives for Guyana. 1973.44 p. Report of workshop with the National Science Research Council of Guyana describes new methods of aquatic weed control suitable for tropical developing countries. Roofing in Developing Countries: Research for New Technolo8ies. 1974. 74 p. (out of print) Emphasizes the need for research on low cost roofs, particularly using materials available in developing countries. NTIS Accession No. PB 234503/AS. $4.75. Meeting the Challenge of Industrialization: A Feasibility Study for an International Industrialization Institute. 1973. 133 p. (out of print) Advances concept of an independent, interdisciplinary research institute to illuminate new policy options confronting all nations. NTIS Accession No. PB 228-348. $5.75.

97

14.

15.

16.

17.

20.

More Water for Arid Lands: Promising Technologies and Research Opportunities. 1974. 153 p. Outlines little-known but promising technologies to supply and conserve wa ter in arid areas. International Development Programs of the Office of the Foreign Secretary, by Harrison Brown and Theresa Tellez. 1973. 68 p. History and analysis, 1963-1972; lists staff/participants and publications. Underexploited Tropical Plants with Promising Economic Value. 1975. 187 p. Describes 36 little-known tropical plants that, with research, could become important cash and food crops in the future. Includes cereals, roots and tubers, vegetables, fruits, oilseeds, forage plants, etc. The Winged Bean: A High Protein Crop for the Tropics. 1975.43 p. Describes a neglected tropical legume from Southeast Asia and Papua-New Guinea that appears to have promise for combatting malnutrition worldwide. Systems Analysis and Operations Research: A Tool for Policy and Program Planning for Developing Countries. 1976.

In Preparation (Working Titles)

5. 18. 19. 21. 22. 23.

The Role of U.S. Engineering Schools in Foreign Assistance Activities. Energy for Rural Development: Renewable Resources and Alternative Technologies for Developing Countries. Methane Generation from Human, Animal, and Agricultural Wastes. Productive Utilization of Freshwater Aquatic Weeds. Guayule: Rubber Producing Desert Shrub. Remote Sensing for Development.

Related Publications Other reports (prepared in cooperation with BOSTID) available from the above address are: An International Centre for Manatee Research. 1975. 34 p. Describes the use of the manatee, a large, almost extinct, marine mammal, to clear aquatic weeds from canals. Proposes a research laboratory to develop manatee reproduction and husbandry. Published by the National Science Research Council of Guyana. Products from Jojoba; a Promising New Crop for Arid Lands. 1975.30 p. Describes the chemistry of the oil obtained from the North American desert shrub Simmondsia chinensis.

How to Order Published reports (unless out of print) are available free on request from the Board on Science and Technology for International Development. BOSTID will fill requests for reports in preparation upon publication. Out-of-print reports are available from the National Technical Information Service. To order, send report title, NTIS Accession Number, and amount indicated. Pay by NTIS Deposit Account, check, or money order. U.S. orders without prepayment are billed within 15 days; a 504 charge is added. Foreign buyers must enclose payment plus U.S. $2.50 handling charge per item. Send order to: National Technical Information Service Springfield, Virginia 22161, USA

98

--

I

- - - - --

-~-

'll

Order Form

I

I

I

Please indicate on the labels below the names of colleagues. institutions, libraries, etc., that might be interested in receiving a copy of Systems

~

I I

, I

I

Analysis and Operations Research: A Tool for Policy and Program Planning for Developing Countries.

Please return this form to

!I

,, I

Commission on International Relations (JH 214) National Academy of Sciences 2101 Constitution Avenue, N.W. Washington, D.C. 20418, USA

I

I

, I

I

:

t

I

I

:

I-----------------r---------------------

,

'

I

I

II

I ,

I ,

I

20

l

20

I ~-------------r-------------

I I

I

!I

I I

I, : I

I I

20

I

20

I

:-- ---- ------------1- -- ------- ------ ------I

I

\

t

I

I

I

I

:

I

I

I

\

I

I

I

20

I.

')n

---I

D

1

C

I

C:oogk

NATIONAL ACADEMIES LIBRARY

08603

1111111 "'" 11111 11111 "'" 1111111'

(. ()o~I

Loading...

J 3tems Analysis and Operations Research - USAID

3tems Analysis and Operations Research: - J 'A TOOL FOR POLICY AND PROGRAM PLANNING FOR DEVELOPING COUNTRIES Board on Science and Technology for In...

4MB Sizes 3 Downloads 0 Views

Recommend Documents

Field Operations Guide - usaid
(where present) with the management of the USG response to a disaster. As with an Assessment Team, DARTs ... other donor

OPTIMIZATION AND OPERATIONS RESEARCH
OPTIMIZATION AND OPERATIONS RESEARCH. 10. Generalizations. 11. The Art of Dynamic Programming. 11.1. Models. 11.2. Algor

Optimization and Operations Research
Optimization: the mathematical theory of models and algorithms. 5.1. Deterministic Optimization. 5.1.1. ... mathematical

2 Operations Research and Intelligence Analysis--Edward H. Kaplan
Read chapter 2 Operations Research and Intelligence Analysis--Edward H. Kaplan: The U.S. intelligence community (IC) is

sector environmental analysis - USAID
Jul 31, 2008 - USAID Bangladesh is committed to environmental sensitivity in all of its activities. .... PRICE SECTOR EN

Operations Research
1.1 THE ORIGINS AND THE NATURE OF OPERATIONS RESEARCH 5 .... solution. • Test the model and refine it as needed. In th

Operations research and statistics pdf
Cyrill prepositional capitulated its staggered eagle. spongy makalah rekayasa genetika pada hewan and diplomatic Jerri p

applied agriculture research project - USAID
Sep 29, 1980 - All conditions precedent for initial disbursement under both loan and grant have been met. 2. A "Detailed

Supporting Research and Analysis
Sep 15, 1999 - integration will nevertheless produce growing inequality within and .... I: Global Dynamics. The future i

Operations Research - Whitman People
See me if you have problems obtaining the text. The ISBN is 0534359647. Alternate Text: The first 10 chapters of this te