absolutely positively operations research: 50 years of

24
1 Draft December 10, 2002 Absolutely Positively Operations Research: 50 Years of Contributions by William Wager Cooper Arnold Reisman* ^ Muhittin Oral* Said Gattoufi* *Graduate School of Management, Sabanci University, Istanbul, Turkey ^Reisman and Associates, Shaker Heights, OH. USA ABSTRACT This paper addresses some of the contributions to real operations research over one lifetime – the case of William Wager Cooper. Key words: Efficiency analysis, Data Envelopment Analysis, Goal Programming.

Upload: others

Post on 11-Jun-2022

0 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Absolutely Positively Operations Research: 50 Years of

1

Draft December 10, 2002

Absolutely Positively Operations Research:

50 Years of Contributions by William Wager Cooper

Arnold Reisman*^ Muhittin Oral* Said Gattoufi*

*Graduate School of Management, Sabanci University, Istanbul, Turkey ^Reisman and Associates, Shaker Heights, OH. USA

ABSTRACT

This paper addresses some of the contributions to real operations research over one lifetime – the case of William Wager Cooper.

Key words: Efficiency analysis, Data Envelopment Analysis, Goal Programming.

Page 2: Absolutely Positively Operations Research: 50 Years of

2

Absolutely Positively Operations Research:

50 Years of Contributions by William Wager Cooper1

1. Introduction

The above title paraphrases that of an article written by Mason et al (1997). Whereas

the Mason paper discussed a specific case history of real operations research in a real

company - FED EX, this paper addresses the contributions to real operations research over

one lifetime – those of William Wager Cooper. After seeing an early draft of another paper

(Reisman and Kirschnick (1995)), Bill Cooper, recalling from his own experience, wrote:

This paper stimulated my thinking and also brought back many memories. One of the possibilities to be considered is the reinforcing effects which may occur when several of the strategies you describe are employed simultaneously. A case in point from my own experience is the original article which Abe Charnes and I wrote with Bob Mellon and published in the April 1952 issue of Econometrica (a really abstract methodology oriented journal) entitled "Blending Aviation Gasolines: A Study in Programming Interdependent Activities in an Integrated Oil Company" (Charnes, Cooper and Mellon 1952). This was the first reported actual application of linear programming and the effect was enormous both on industrial practice in more than one industry, and theoretical-methodological research (in more than one discipline). Many things were involved-a new application, new methodologies and new substantive theory. Perhaps this was due to the mix of disciplines in our team which included chemical engineering and refinery experience (Mellon), mathematics and engineering (Charnes) and economics, management and accounting (Cooper (1994)).

At this point, Cooper inserted a footnote that, "We only discovered at a later date that

this was to be called 'operations research' or still later, 'management science’”. He then

continued:

These wide ranging and continuing effects, or at least the speed which these occurred, may also have been due to the times and the psychological aftermath (of euphoria) resulting from the 'great historical divide' we now refer to as 'World War II”. (Cooper 1994).

A creative application2 of linear programming followed the structuring of blending

gasoline into a mathematical model in the above example. The results were empirically 1 The focus on a single individual’s contributions to the literature of OR/MS represents an extension of meta research performed on OR/MS in general (Reisman and Kirschnick, (1994, 1995 and 2001)); and on several OR/MS sub-disciplines (Reisman et al. (1997a and b), and (2001)), and (Gatouffi et al., 2001a,b,c, and d.)). 2 Italicized descriptors refer to research strategies described in Reisman (1988) and (1992) and applied in

Page 3: Absolutely Positively Operations Research: 50 Years of

3

validated by industrial practice. The work involved new models, new methods and

substantive theory as a result of bridging the state of knowledge of linear programming and

that of chemical engineering and bringing the results to bear on the managerial problems

that were addressed. It provides a good example of what Cooper was later to refer to as

“Applications driven theory” (Cooper and Mc Allister (1999))

The above article, as co-authored by Cooper, and the DEA literature as a whole leave

“absolutely positively” no doubt that a true interdisciplinary approach to real-life problem

solutions can yield major contributions to practice and, yes, to theoretical developments as

well. This conclusion was reached not by looking at a single article describing a single

development at a single point in time, but by reviewing most of Cooper’s 22 books and 475

publications in refereed journals over his 50 year career. Clearly there are others such as

Abe Charnes, Hugh Miser, and T. C. Koopmans who over their lifetime made major

contributions to theory as the result of work in structuring and solving real world problems.

2. Goal Programming

Goal Programming was first introduced by Charnes and Cooper in 1961 and its literature

has grown by leaps and bounds ever since. The title of a recent article “Goal programming

model: A glorious history and a promising future” by Aouni and Kettani (2001) tells it all.

However to elaborate we quote:

Who would have expected it? Who would have predicted that goal programming (GP) introduced by Charnes and Cooper in the early 1960s as a simple linear program, would have leaped to such success in the 21st century? Far from being a timid or hesitant leap of success it is instead a promising jump marked with confidence. Today GP is alive more than ever, supported by a network of researchers and practitioners continually feeding it with theoretical developments and applications, all of these with resounding success. In fact, GP has hundreds of monographs and scientific papers in its favor, and hundreds of applications covering an impressive number of areas and disciplines. As a bonus, the Internet has also joined in with the emergence of WEB sites and discussion groups. Aouni and Kettani (2001)

Reisman and Kirschnick (1995) and in Reisman et al (1997a and b)

Page 4: Absolutely Positively Operations Research: 50 Years of

4

Its theoretical and applied literature base has been reviewed at several points in time by

many including Zanakis and Gupta (1985), Romero (1986, and 1991), Ignizio and Cavalier

(1994), Schniederjans (1995), and Tamiz et al. (1998). Gabriel Tavares at Rutgers University

has compiled a GP bibliography with over 3200 entries. ([email protected]) And, this is

how Bill Cooper recollects the events:

Goal programming was developed with Bob Ferguson. A member of the staff of Methods Engineering Council (MEC), a Pittsburgh based consulting firm, Bob was engaged in helping them to develop guides for determining the executive compensation plans for use by the Major Industrial Appliances Division of the General Electric Co. After completing very extensive (and expensive) surveys and interviews,3 MEC began to try to develop the desired executive compensation scheme by applying these data to various types of statistical regressions. Difficulty was experienced by MEC, however, because these statistical regression approaches all gave unacceptable results. Coefficients with the wrong sign and failure to reflect GE’s organization hierarchy in its salary estimates were some of the sources of trouble.4 We pointed out to Ferguson that a reflection of these conditions would naturally lead to inequality rather than the equation formulations used in ordinary statistical regression estimating models.

Also important was our noting that uses of company salary records to obtain these regression estimates failed to respond to the problem that concerned the company -- viz., to meet salary offers from competitors which were designed to attract valuable company personnel away from GE. The objective of meeting such competitors was to be accomplished, however, while conforming to the company’s organization constraints and policies. Access to the data associated with such competing salary offers was not available, of course, but we could at least obtain estimates of upper and lower bounds for their values. This, too, lent itself to the use of linear inequalities. However, the objective “meet competing offers as closely as possible” was given an absolute value function (nonlinear) formulation in the objectives of the models we formulated in the course of successive meetings with Ferguson. Algorithms for such objectives were not available. However, they were also not needed because we were able to show how such formulations could be reduced to equivalent linear programming problems for which, by this time, computer codes as well as algorithms were already available. To handle the problem of computations, we secured the services of Alex Ordon who had left Dantzig’s organization in the U.S. Air Force to join the Burroughs Corporation in its efforts to move from making mechanical adding machines into electronic computers.

This modeling effort, to estimate executive salaries for guidance to GE management, led to the development of a new approach to absolute value regressions (under constraints) which we referred to as “inequality constrained regressions.” This helped to appeal to the statistics literature -- for which we might cite the treatment of absolute value regressions in G. Bassett, Jr., and R. Koenker (1978). See also the appendix to R. Koenker and G. Bassett, Jr. (1978)

3 The idea was to determine “job factors” and “man factors” which could enter as weights in determining the compensation of individual executives and to do so in a manner that could provide incentives to each executive which could help him improve his future capabilities for the company. This was a novel idea in job description approaches in that it recognized that executives (unlike factory workers) could change their job descriptions as well as their personal characteristics. We were later able to extend these ideas for work we did with R. Niehaus for use in the civilian manpower planning work we did for the U.S. Navy. These extensions allowed for interactions between changing personal characteristics and the job that could be performed--as needed for planning changes in organization designs that could deal more adequately with improving the equal employment opportunities that could then be provided. See Charnes et al., (1972, 1978). 4 Uses of ordinary (unconstrained) regressions, as had been done by MEC, led to anomalies like the ones reflected in the following comments by the president of this GE Division who is reported to have said: “I am willing to admit that the “office boy” is smarter than I am [as reflected in his “man-factor,” “job-factor” weights] but I am not willing to give him a salary higher than mine [as your formulas might suggest]!”

Page 5: Absolutely Positively Operations Research: 50 Years of

5

and the following quotation from Dielman and Pfaffenberger (1982, p.32): Until Charnes, Cooper and Ferguson (1955) demonstrated that LAV [Least absolute Value] estimators could be produced by linear programming methods, it was simply not feasible to consider [using] the LAV estimator in most applications…. This development of “inequality constrained regression,” with its attendant relations to

linear programming stimulated, in turn, a whole series of developments which included replacing earlier (calculus based) optimization methods and proofs with more general mathematical programming reformulations and extensions to statistics--many of which are represented, along with novel developments of their own, in the remarkable book by Arthanari and Dodge (1981), which shows how virtually all of the commonly used statistical methods --- ranging from regression to cluster analysis --- can be treated by variations of these methods. .

These approaches also provided a new way of dealing with inconsistencies in linear programming that could be brought to bear on programming to meet “multiple objectives” “as closely as possible.” This represented a new class of problems which we were encountering with increasing frequency in our work, and this supplied the motivation for our choice of the name “goal programming” -- which made its initial appearance in Appendix B of A. Charnes and W.W. Cooper (1961).

In the process of adapting these methods to other problems, it was possible to combine

these goal programming approaches with other approaches in ways that greatly enhanced their

power for use in new applications. The work that Cooper did with A. Charnes and R. Niehaus to

develop ways to deal with problems the U.S. Navy encountered in implementing its equal

employment opportunity policies is a good example. To deal with the dynamics that formed an

essential feature of these problems goal programming was combined with Markoff processes in a

series of models that they referred to as ''goal programming with embedded Markoff processes.''

(A. Charnes, W. W. Cooper, K.A. Lewis and R.J. Niehaus (1978))

These models also proved useful in addressing problems other than manpower planning such as

the introduction of new organization elements by the Navy in the form of ''bridge positions''.

These opened new paths to enable minorities entering in clerical and stenographic positions to

subsequently move over into professional, technical and supervisory career paths with associated

probabilities. These models also proved useful in several lawsuits which had been brought

against the Navy by enabling the Navy lawyers to distinguish this use of ''goals'' from the use of

quotas'' in EEO planning. Finally they helped to avoid the issuance of a manpower policy (for

the entire Federal Government) which was to take form in a proposed U.S. Office of

Management and Budget Circular. By use of these models it was possible to show that the

proposed policy would lead to contradictions that have made its implementation undesirable.

Page 6: Absolutely Positively Operations Research: 50 Years of

6

We now turn to yet another major contribution to OR/MS with a W.W. Cooper imprint.

3. Data Envelopment Analysis

DEA was first introduced by Charnes, Cooper, and Rhodes. (CCR) in 19785. As is

shown in Figure 1, its literature has grown exponentially so that by August of 2001, a total of

1797 articles are known to have been published in 490 different refereed journals worldwide.

Many of these publish in languages other than English. The list of refereed journals having

DEA content includes those publishing in French, German, Italian, Spanish as well as a number

of languages from East Europe, Scandinavia, Asia/Pacific, and the Near East indicates the

field’s growth and diffusion. To obtain a more in-depth view, its life-cycle literature was

reviewed and classified (Gattouffi et al., 2001d) on a scale ranging from pure theory to bona

fide application. Additionally, 989 of its papers were classified in terms of the seven types of

research processes suggested in Reisman (1988,1992) used by their authors. Lastly, the DEA

results were compared with those similarly obtained for other OR/MS subdisciplines e.g.,

Flowshop Scheduling and Sequencing (FSS), Cellular Manufacturing (CM) and Game Theory

(GT) respectively discussed in Reisman et al. (1997a, 1997b and 2001). Based on that meta

research Gattouffi et al., (2001d) conclude:

“Anemia in relevance to the real world”, an important symptom of what has variously been called a “natural drift” toward academization of a discipline (Abbott (1988), Corbett, and Van Wassenhove (1993)), away from the “swamps of relevance” (Miser, 1987), would be characterized by a trend toward a preoccupation with extensions of theory, albeit small, and or excessive use of the “ripple” strategy6 especially in …. articles which Ormerod and Kiossis (1997) call “untested theory”. Over time, the DEA literature is getting ever more grounded in the real-world and is continuously identifying and indeed working the new “swamps of relevance”. ……

Clearly DEA is not heading the way of U.S. based academic OR/MS in general (Reisman and Kirschnick 1994, 1995). It is bucking the “natural drift”, and perhaps helping to resurrect OR/MS from the “post mortem” state suggested by Ackoff (1987).

5 It must be acknowledged that Farrell (1957) and Boles (1971) present many concepts upon which DEA is based as discussed in Forsund and Sarafoglou (2002). 6 One of the seven distinguishable research strategies defined in (Reisman 1988, 1992)

Page 7: Absolutely Positively Operations Research: 50 Years of

7

Because of the limitation of the GT7 sample used, caution is indicated in the above comparative analysis between GT and DEA. However, significant differences between the DEA literature and those of CM and FSS are confirmed by the statistical results reported….

DEAs relative vitality is confirmed by the higher “compounding” rate in the accumulation of its literature compared to FSS and CM, (Gattoufi et al 2001c). The diffusion of DEA to other disciplines and professions is indicated by the ever increasing number of journals, in turn representing an ever increasing diversity of mission and of readership.

[C]ompared to FSS and CM, DEAs authors show greater creativity8 in their pursuit of new knowledge.

Finally, although the bridging between DEA and GT is very well established, not much was done to establish such connections with the other two subdisciplines. DEA can well provide hope for these sub-disciplines to become tools of OR/MS practice. Thus far, only a single paper combining DEA with CM was identified by the authors within the DEA literature. (Gattouffi et al 2001d)

7 The sample used was limited to all GT papers that have ever been published in Operations Research, Management Science, and Interfaces. Clearly, the GT literature is not so limited. 8 Based on the mix of research strategies (Reisman 1988, 1992) invoked by the authors

Page 8: Absolutely Positively Operations Research: 50 Years of

8

Figure 1: Cumulative Number of DEA Articles for the Period 1978-2001 (Logarithmic Scale)

Reproduced from Gattoufi et. al. (2001c)

In 1987, Blumstein called on the OR/MS community to do more "missionary" type work

addressing the plethora of social problems facing society. This view was supported by the NSF9

funded CONDOR (1988) project. "Like most fields, if theory is not stimulated by practice OR

can become stale and fail to contribute to social needs. Therefore the field must address

problems of practical importance".

From its very outset and throughout its lifespan, DEA articles reported real and often

creative applications1011 in the evaluation of; educational programs (Charnes, Cooper and

Rhodes (1978)); academic departments in Israel (Sinuany-Stern et al. (1994)); universities in

Australia (Avkiran (2001)); courts and criminal justice systems (Lewin et al. (1982)); other

agencies of government (MacMilan (1987)) nursing services (Nunamaker (1983)); hospitals

(Grosskopf and Valdamanis (1987)); health service systems (Charnes et al. 1988); not for profit

organizations (Nunamaker (1985)) and (Charnes and Cooper (1980)); the state of society as

measured by multiple social indicators (Hashimoto and Ishikawa (1993)), environmental

9 National Science Foundation. 10 Another of the seven distinguishable research strategies defined in (Reisman 1988, 1992) 11 This is also the case for GP although the field had not been similarly studied.

1

10

100

1000

10000

1978

1979

1980

1981

1982

1983

1984

1985

1986

1987

1988

1989

1990

1991

1992

1993

1994

1995

1996

1997

1998

1999

2000

2001

Years

Cum

ulat

ive

Num

ber o

f Pub

licat

ions

(Log

arith

mic

)

Page 9: Absolutely Positively Operations Research: 50 Years of

9

controls in the power industry (Fare et al. (1986)); performance in branch banking (Oral and

Yolalan (1990)); quality of hospitality services (Haywood (1983)); road construction in Norway,

(Odeck (2001)); and the list goes on.

Starting in the late 1940s and throughout his 50-year career, W. W. Cooper did

precisely what Blumstein called for and the CONDOR committee composed of "24

operations professionals representing 15 U.S. universities, one foreign university and three

U.S. companies" indicated. His work was obviously noticed and appreciated – by many.

Dirkmaat (2001) reports that among the most cited papers published in the Journal of

Econometrics, the Seiford and Thrall (1990) DEA related paper ranks 14th with 121 citations

and the Charnes et al. (1985) DEA related paper ranks 18th with 102 citations. This shows

the high significance of DEA for researchers far and beyond the OR/MS community. We

now turn to a review of the literature.

3.1 Diffusion of DEA

Through content analysis of management publications, Corbett, and Van

Wassenhove (1993) noted the declining number of articles dealing with OR/MS over the

years 1956-1991 in such management journals as the Harvard Business Review. Based on

such findings they concluded that OR/MS was being overtaken by a “natural drift” away

from what Miser (1987) referred to as the “swamp of relevance”. One approach to

measuring the diffusion of DEA from its OR/MS home base and back into the “swamp of

relevance” is to once again, look at the publishing patterns and outlets of choice used by

authors writing on DEA and to compare this with other OR/MS subdisciplines and OR/MS

as a whole. As indicated this was indeed done and the following findings again reconfirm

the relevance of DEA to a world beyond the OR/MS establishment’s bounds. Such

differences can be explained by one of at least two hepotheses.

The first hypothesis is that the U.S. based OR/MS community/establishment was “asleep

at the wheel” so to speak or worse yet was operating on a paradigm that was out of sync with

the field’s original paradigm. Instead of working “in the swamps of relevance” applying

Page 10: Absolutely Positively Operations Research: 50 Years of

10

“whatever tools they had” (Miser (1987)) in order to solve problems of “significance to

society” (Blumstein (1987)), the establishment became preoccupied with improvements, no

matter how incremental or minute, to the extant theoretical base (Reisman and Kirschnick

(1994 and 1995)). In saying that “[m]any of our educational programs--and some regarded as

the very best”, Pierskalla (1987), implied that this unfortunate paradigm dominated the

American OR/MS academic scene as far back as the nineteen eighties. Sadly, to a large extent,

it still does.

A possible source of these differences is the fact that in its original paradigm OR/MS

was and is concerned with “planning” decisions and with “forecasting” while DEA is

essentially an ex post facto evaluation for the “control aspects of management” (Cooper

(1999)). Consequently, DEA is out of sync with the existing OR/MS establishment’s paradigm

on two counts.

This hypothesis is further supported by the fact that nine out of the ten most prolific

contributors to the DEA literature are associated primarily with North American

institutions – one is a Canadian - and, yet the European Journal of Operational Research

alone, with a count of 204, published more DEA articles than the total of those published

in the US based flagship journals. The latter currently stands at 67. This was true in the

early DEA years and has been true ever since. Significantly, even though the founders of

the field, Charnes, Cooper and Rhodes, were all U.S. based, the first article in the field was

published in a European journal.

The other hypothesis is that significant diffusion into other disciplines and

professions has taken place. Supporting this is the fact that currently DEA articles can be

found in 490 refereed journals worldwide. Many of them publish in languages other than

English. The list of refereed journals having DEA content includes those publishing in

French, German, Italian, Spanish as well as a number of languages from East Europe,

Scandinavia, Asia/Pacific, and the Near East. These journals are quite varied in many

other ways. This will be demonstrated in several ways.

Table 3, lists the top thirty journals ranked in order of DEA article counts. The titles

of the journals listed further demonstrate the diffusion of DEA. Moreover its diffusion

Page 11: Absolutely Positively Operations Research: 50 Years of

11

follows each of two dimensions. One dimension represents diffusion across disciplines and

professions, the other across national boundaries – indeed across continents. A large

percentage of refereed journals publishing DEA subject matter are not U. S. based.12

Table 3: Top 30 Journals ranked in terms of the number of DEA publications and the number of DEA articles they have published

Reproduced from Gattoufi et. al. (2001c)

12Many journals have an editor-in-chief headquartered in one country and the publishing organization in another, as is the case with all the Pergammon and Elsevier journals. Moreover some have been known to move the editor-in-chief headquarters across continents as has recently been the case with Omega. Consequently, it is difficult to get an exact count of non-US based journals.

Page 12: Absolutely Positively Operations Research: 50 Years of

12

Rank Journal Total Number of DEA-Articles the Journal Published

1 European Journal of Operational Research 2042 Journal of Productivity Analysis 1093 Journal of The Operational Research Society 774 Annals of Operations Research 535 Management Science 516 OMEGA 507 Applied Economics 428 Socio-Economic Planning Sciences 399 International Journal of Systems Science 37

10 International Journal of Production Economics 3311 Computers & Operations Research 2512 Journal of Banking & Finance 2513 Journal of Econometrics 1914 Journal of The Operational Research Society of Japan 1815 Applied Economics Letters 1716 Managerial & Decision Economics 1617 Review of Economics and Statistics 1618 American Journal of Agricultural Economics 1519 INFOR 1420 Journal of Medical Systems 1421 Health Care Management Science 1322 Research in Governmental & Nonprofit Accounting 1223 INTERFACES 1124 Medical Care 1125 Transportation Research 1126 Health Services Research 1027 Computers, Environment & Urban Systems 928 Operations Research Letters 9

29 Operations Research: Communication of the Operational Research Society of Japan 9

30 Decision Sciences 8

There is yet another way to measure the diffusion of DEA out of OR/MS. Of the 490

refereed journals having DEA articles only ten (10) (slightly over 0.2%) can be considered

to be hard core OR/MS journals13. As is shown in Gattoufi et al. (2001c) they account for

28 % of the DEA literature extant August 2001. Thus, almost three-quarters or 72% of the

DEA literature was published in archival journals outside its birth discipline. Additional

indicators of DEA’s diffusion to other disciplines is the fact that the Journal of Productivity

Analysis, ranked second in terms of DEA publications, is not an OR/MS journal but a

microeconomic oriented journal and that Applied Economics ranks 7th, Socio-Economic

Planning Sciences ranks 8th,the Journal of Banking and Finance ranks 12th and Journal of

Econometrics ranks 13th.

13 Cited in the order of DEA article count: European Journal of Operational Research, Journal of the Operational Research Society, Annals of Operations Research, Management Science, OMEGA, Computers and Operations Research, INFOR, Interfaces, Operations Research Letters, Operations Research, International Transactions on Operations Research

Page 13: Absolutely Positively Operations Research: 50 Years of

13

3.2 Vitality of DEA

It may well be argued that the vitality of a field of knowledge is not independent of its

diffusion into other disciplines and professions. Be that as it may. The following statistics

provide indications of DEA’ s vitality vis a vis other and older OR/MS subdisciplines e.g.,

FSS, and CM (Gattouffi et al. (2001c)).

1. The accumulation of the literature in each of the three disciplines is shown (at a

high precision of fit) to be exponential with DEA’s growth parameter being 0.255

while that of FSS and CM respectively are 0.151 and 0.106.

2. This is analogous to a 25.5% rate of interest, compounded annually, on a money

deposit versus one of 15.1% or 10.6%.

3. The official year of birth for FSS is 1952, for CM it is 1969 and for DEA, it is 1978.

Thus DEA is by far the youngest of the three disciplines.

4. The total number of FSS, CM and DEA papers published in refereed journals circa

August 2001, is respectively; 316, 374, and 1797.

5. Thus the average (over the discipline’s lifetime) number of papers published per

year is, respectively, 316/49 = 6.4, 374/32 = 11.7, and 1797/23 = 78.2

Page 14: Absolutely Positively Operations Research: 50 Years of

14

Figure 2: Cumulative Number of Papers Published in Refereed Journals for DEA,

CM and FSS Plotted in a Semi-Logarithmic Scale Reproduced from Gattoufi et. al. (2001c)

DEA is by far the youngest of the three disciplines. Figure 2 clearly shows that during each

year of its lifecycle, DEA has never fallen below the accumulated literature of the other two

disciplines’ corresponding life-cycle year. It is almost an order of magnitude greater than those of

the much older disciplines. Moreover it experienced the shortest gestation period14, e.g., the time

until the plot of its time series approximates a line on semilog coordinates. The gestation period for

FSS appears to be 7 years, or 16 percent of its lifetime. The gestation period for CM appears to be

4 years, or 12 percent of its lifetime while the gestation period for DEA appears to be 2 years, or 8.3

percent of its lifetime. (Gattouffi et al (2001c)).This implies that within its lifetime DEA

experienced earlier acceptance and much greater interest than did the other two disciplines. This is

yet another argument for DEA’s comparative vitality.

After seeing an early draft of this paper Finn R. Forsund (2002), a major contributor to the

14 The time expired between publication of the field’s launch article and the beginning of the takeoff stage in its literature.

1 6 11 16 21 26 31 36 41 46

Year of the life cycle

Cum

ulat

ive

Num

ber o

f Pap

ers Data Envelopment Analysis

Cellular Manufcaturing

Flowshop Scheduling

10

100

1000

Page 15: Absolutely Positively Operations Research: 50 Years of

15

DEA literature, wrote:

Cooper’s interest in and inspirations from real life applications reminded me of Ragnar Frisch. As an example I can take a story he [Frisch] told about an engineer/economist, who studied coal production. [Frisch] is full of praise for economists taking the trouble of really coming to grips with real production processes. According to Frisch (1970), in his paper presentation the engineer /economist spent more than one-third of his time explaining what a coal pit is and what the profile of a coal pit is. [The audience] sort of felt that his paper was written with dirty fingers because he had just come back from digging in the coal pits. From this concrete preoccupation he derived his theoretical concepts and formulated his programming problem which then appeared as a problem full of life and reality.

The paper by Charnes, Cooper and Mellon (1952), which launched Goal Programming as a

field was clearly written on paper that reeked with petrochemicals. Charnes, A., Cooper, W.W.,

and Rhodes, E. (1978), commonly accepted as the launch of Data Envelopment Analysis,

without any doubt, still resounds with the noise that is common in the hallways of public schools

in America.

Bill Cooper is widely acknowledged to be among the grand old men of OR/MS. Yet, his

early contributions to the economics profession in the 1950s led to his election as Fellow of the

Econometric Society. Similarly, for his contribution to accountancy he was accorded the

Outstanding Accounting Educator Award by the American Accounting Association, and elected

to the Accounting Hall of Fame. Of course, he is a recepient of the John von Neumann Theory

Prize from what was then known as TIMS/ORSA. The title of a recent (2002) paper appearing

in the Journal of Productivity Analysis, by S.C.Ray, William W. Cooper: A legend in His

Own Times, says it all. In no way is this a eulogy for W. W. Cooper. Unlike Charnes, Miser

and Koopmans, he is still with us and his contributions to both theory and to the practice of

OR/MS continue unabated.

4. Concluding Remarks

Irrespective of whether the OR/MS establishment during the 1980’s and 1990’s recognized

DEA to be “absolutely positively operations research”, the world around it was in need of a better

Page 16: Absolutely Positively Operations Research: 50 Years of

16

tool for evaluating the performance of organizations and the demonstrated embracing of this

methodology speaks volumes about its utility.

Over his 50+ year career William Wager Cooper has been totally unaffected by the very

significant “natural drift” away from the “swamps of relevance” and from “missionary work”

toward “introversion”, “loss of relevance”, “devolution”, and “mechanical optimi[zation]”, which

took place during that same time-frame among the OR/MS academic establishment in the United

States. History has borne out that W. W. Cooper was correct in keeping his course firmly rooted in

the very “swamps of relevance” while significantly and meaningfully extending and expanding the

theoretical basis of OR and of MS, giving other professions a sought after tool and thus enabling the

kind of “missionary work” that Blumstein called for.

Specifically, DEA appears to provide hope for Game Theory to be more real-world

friendly in the future. Bridging of the two fields started during the early days of DEA and a

large literature exists that relates GT to DEA. If these efforts succeed then the award of the

Nobel Prize for GT will have been visionary and even more so justified. If such is the

outcome , W. W. Cooper will have played a major role in its coming about.

We conclude with a few brief remarks about other aspects of Cooper's activities and

their effects on both theory and practice . In addition to his numerous (over 400 articles

and 22 books) publications, Cooper produced a large number of graduates including

nearly 100 PhDs who have gone on to careers in research, teaching and practice. The

effects of his teaching activities have gone beyond the classroom and taken such forms as

preparation of texts, cases and other types of teaching materials. According to a recently

published short resume: “He has served as consultant to more than 200 business firms and

government agencies around the world” (About the Authors (2002)). Having at all times

kept one foot in the real world has enriched his research, brought realism into his

classrooms and to his guidance of doctoral research. This in turn, has made him a sought

after management consultant.

In 2002, commerorating the “50th anniversary of the founding of ORSA”, Saul Gass

included not one but two of the books coauthored by W. W. Cooper among “OR’s Top

25: Twenty five books that shaped how operations research is taught and practiced.” This

includes the 1953 text coauthored with Abe Charnes and A. Henderson, “An Introduction

Page 17: Absolutely Positively Operations Research: 50 Years of

17

to Linear Programming” and the 1965 text Management Models and Industrial

Applications of Linear Programming which Cooper authored with A. Charnes.

The OR profession owes much to these …authors’…insights, research, writing skills, and devotion to their craft. Gass (2002)

Indeed it does! With Cooper, this kind of activity continues to the very present.

Until recently, DEA texts like Fare, Grosskopf and Lovell (1985, 1994) were heavily

oriented toward academic audiences in economics. However, the year 2000 text by

Cooper, Seiford and Tone, notes in its preface that it is oriented toward practitioners as

well as for classroom use. The following letter, made available to us by Cooper, illustrates

some of what can be accomplished with such texts:

I'm writing to say thank you for the text "Data Envelopment Analysis." I have found it very helpful in understanding this rather powerful tool.

Lipper is a mutual rating/ranking service similar to Morningstar, though our clients tend to be institutional based (pension funds, advisors and the fund companies themselves) versus individuals. The ability to combine various inputs and outputs related to funds and come up with an overall "score" is something we are finding very helpful. This is especially true in the board reporting segment where we present to the boards of fund companies or plan sponsors how their funds are doing versus their competitors or other funds.

We are also looking into using DEA as a peer group selection tool, which is another major business for us.

Again, many thanks for writing such a terrific text.

Andrew Clark Senior Research Analyst Lipper, A Reuters Company

References

Abbott, A., (1988). The Systems of Professions: An Essay on the Division of

Expert Labour, University of Chicago Press, Chicago

About the Authors (2002). Socio economic Planning Sciences. 36 (4), 309

Page 18: Absolutely Positively Operations Research: 50 Years of

18

Ackoff, R. (1979) The Future of Operations Research is Past. Journal of the Operational

Research Society, 30, 93-104.

Ackoff, R. (1987) OR: A Post Mortem. Operations Research 35, 471-474.

Aouni B.and Kettani O. (2001) Goal programming model: A glorious history and a

promising future. European Journal of Operational Research; 133, 225-231. Special

Issue on Multi Objective Programming and Goal Programming.

Arthanari, T.S. and Y. Dodge (1981) Mathematical Programming in Statistics (New

York: John Wiley & Sons).

Avkiran, N. K.(2001) Investigating Technical and Scale Efficiencies of Australian

Universities through Data Envelopment Analysis. Socio-Economic Planning

Sciences; 35 (1): 57-80.

Avkiran, N. K., (2002) Productivity analysis in the services sector with data envelopment

analysis. 2nd Edition, University of Queensland at Ipswich Publications, Ipswitch,

Australia.

Basset, G., Jr., and R. Koenker (1978) “Asymptotic Theory of Least Absolute Value Error

Regressions,” Journal of the American Statistical Association 73, pp. 618-622.

Blumstein, A. (1987) The current Missionary role of OR/MS Operations Research 35,

926-929.

Boles, J. N. (1971). The 1130 Farrell Efficiency System-Multiple Products, Multiple

Factors. Giannini Foundation of Agricultural Economics, February. See Forsund,

and Sarafoglou, 2002.

Charnes, A. and Cooper W. W. and Mellon B. (1952) Blending Aviation Gasolines – A

study in Programming Interdependent Activities in an Integrated Oil Company.

Econometrica, 20 (2), 135-159.

Charnes, A. and Cooper W. W. and Ferguson (1955) Least Absolute Value

Charnes, A., W.W. Cooper, K.A. Lewis and R.J. Niehaus (1972) Studies in Manpower

Planning (Washington, D.C.: U.S. Navy, Office of Civilian Manpower

Management).

Charnes, A., W.W. Cooper, K. A. Lewis and R. J. Niehaus (1978) Equal employment

opportunity planning and staffing models. In D. T. Bryant and R J. Niehaus.

Page 19: Absolutely Positively Operations Research: 50 Years of

19

Manpower Planning and Organization Design, Plenum Press, New York.

Charnes, A., W.W. Cooper, R.J. Niehaus and A. Stedry (1978) “Static and Dynamic

Assignment Models with Multiple Objectives, and Implications for Organization

Design,” Management Science 15, B365-B375.

Charnes, A. and Cooper W. W. (1980) Auditing and accounting for program efficiency and

management efficiency in not-for-profit entities. Accounting, Organizations and

Society 5(1):87-107.

Charnes, A. and Cooper W. W., Henderson, A. (1953), An Introduction to Linear

Programming. Wiley, New York.

Charnes, A. and Cooper W.W. (1961, 1965), Management Models and Industrial

Applications of Linear Programming. Wiley, New York.

Charnes A, Cooper W. W., Rhodes E. (1978) Measuring the Efficiency of Decision

Making Units. European Journal of Operational Research; 2: 429-44.

Charnes A, Cooper W. W., Golany B, Seiford LM, Stutz J. (1985). Foundations of Data

Envelopment Analysis for Pareto-Koopmans Efficient Empirical Production

Functions. Journal of Econometrics; 30: 91-107.

Charnes A., Cooper WW, Assad MJ, Golany B, and Wiggins PE, (1988). Efficiency

analysis of uses of medical care resources in the U.S. army health services command.

In: Kuhn RL (ed). Frontiers of Medical Information Sciences. Praeger, Inc.: New

York.

Charnes A, Cooper W. W., (1981) Rhodes E. Evaluating Program and Managerial

Efficiency: An Application of Data Envelopment Analysis to Program Follow

Through. Management Science; 27(6): 668-97.

CONDOR (Committee on the Next Decade in Operations Research) (1988) Operations

Research: The Next Decade. Operations Research 36, 619-637

Corbett, C. J., Van Wassenhove, L. N., (1993). The Natural Drift: What Happened to

Operations Research?. Operations Research 41, 625-640. Cooper, W. W. (1994) Personal communication.

Page 20: Absolutely Positively Operations Research: 50 Years of

20

Cooper W. W. (1999), Operations Research/Management Science: Where it’s been. Where

it should be going?, Journal of the Operational Research Society. 50. 3-11.

Cooper W. W (2002), Abraham Charnes and W.W. Cooper (et al.): A Brief History of a

Long Collaboration in Developing Industrial Uses of Linear Programming, Invited

article in 50th Anniversary issue Operations Research, 50 (1), 1-7.

Cooper, W. W., Seiford L, Tone K., (2000) Data Envelopment Analysis: A Comprehensive

Text with Models, Applications, References and DEA Solver Software. Kluwer

Academic Publishers, Norwell, Mass.

Cooper, W. W. and L. Mc Allister (1999), Can Research be Basic and Applied? You Bet It

Better Be for B Schools, Socio-Economic Planning Sciences. 33: 257-276

Corbett, C. J and Van Wassenhove L. N. (1993), The Natural Drift: What Happened to

Operations Research? Operations Research. 41, 625-640

Dielman, T. and R. Pfaffenberger (1982) “LAV (Least Absolute Value) Estimation in

Linear Regression: A Review” in TIMS Studies in the Management Sciences

(Amsterdam: North Holland Publishing Co.).

Dirkmaat J. (2001) Some publishing facts, Figures, and observations on the occasion of

Volume 100, number 1 of the Journal of Econometrics. Journal of Econometrics;

100 99-112

Fare R. , and C. A. K. Lovell (1978), Measuring Technical Efficiency in Production.

Journal of Economic Theory, 19, 150-162.

Fare R. , Grosskopf, S. and Pasurka (1986) Effects on Relative Efficiency in Electric

Power Generation Environmental Controls. Resources and Energy 8 (2): 167-84

Fare R. , Grosskopf, S. and Lovell, C.A.K. (1994), Production Frontiers, Cambridge

University Press, Cambridge

Farrell, M. J. (1957) The measurement of productive efficiency, Journal of the Royal

Statistical Society, 125, 252-267

Forsund, F.R., and Sarafoglou, N. 1999. The evolution of DEA-the economics

perspective. Paper presented at the Sixth European Workshop on Efficiency and

Productivity Analysis, Copenhagen.

Forsund, F.R., and Sarafoglou, N. 2002. On the origins of data envelopment analysis.

Journal of Productivity Analysis 17, 23-40.

Page 21: Absolutely Positively Operations Research: 50 Years of

21

Forsund, F. R. (2002), Personal communication.

Frisch, A. (1970), “Econometrics in the world of today”, in Frisch, A. Scott, M. F.G. and

J.N. Wolfe, eds., Induction, Growth, and Trede. Oxford, Clarendon Press,152-166

Gass, S.I. (2002), OR’s Top 25: Twenty five books that shaped how operations research

is taught and practiced, ORMS Today, 29 (4), 40-45

Gattoufi, S., Oral, M., and Reisman, A. (2001a) “Data Envelopment Analysis Literature:

A Bibliography Update (1996-2001)” Working Paper SUGSM – 02-08, Sabanci

University, Istanbul, Turkey. Forthcoming in the Journal of Socio Economic Planning

Sciences

Gattoufi, S., Oral, M., and Reisman, A. (2001b) “A Taxonomy for Data Envelopment

Analysis” Working Paper SUGSM – 02-15, Sabanci University, Istanbul, Turkey.

Forthcoming in the Journal of Socio Economic Planning Sciences

Gattoufi, S., Oral, M., Reisman, A., (2001c). Epistemology of Data Envelopment Analysis

and its Comparison with that of Other OR/MS Fields, Working Paper SUGSM – 02-

14, Sabanci University, Istanbul, Turkey. Forthcoming in the Journal of Socio

Economic Planning Sciences Gattoufi, S., Oral, M., Reisman, A., (2001d). Content Analysis of Data Envelopment

Analysis Literature and its Comparison with that of Other OR/MS Fields, Working Paper

SUGSM – 02-22, Sabanci University, Istanbul, Turkey.

Grosskopf, S. and Valdamanis, V. (1987) Measuring Hospital Performance: a Non

Parametric Approach. Journal of Health Economics. 6; 89-107

Hashimoto, A and Ishikawa, H (1993) Using DEA to Evaluate the State of Society as

Measured by Multiple Social Indicators. Socio-Economic Planning Sciences. 27

(4): 257-68

Haywood, KM (1983)) Assessing the quality of hospitality services. International

Journal of Hospitality Management, 2 (4), 165-177

Ignizio JP (1978). A review of goal programming: a tool for multi objective analysis.

Journal of Operational Research Society. 29, 1109-1119

Ignizio JP and Cavalier TM (1994) Linear Programming. Prentice-Hall, Englewood Cliffs,

Page 22: Absolutely Positively Operations Research: 50 Years of

22

NJ

Lewin, A. Y. Morey R.C. and Cook T.J. (1982) Evaluating the administrative efficiency of

courts. Omega, 10(4):401-411

MacMilan W.D (1987) The measurement of efficiency in multiunit public services.

Environment and Planning. A. 19, 1511-1524

Mason R.O., McKenney J.L, Carlson W and Copeland D. (1997) Absolutely positively

operations research: The Federal Express Story. Interfaces 27: 17-36

Miser, H. (1987) Science and Professionalism in Operations Research. Operations

Research. 35, 314-319.

Nunamaker, T.R. (1983) Measuring routine nursing service efficiency: a comparison of

cost per patient day and data envelopment analysis models. Health Services

Research 18, (2, part1): 183-205

Nunamaker, T.R. (1985) Using data envelopment analysis to measure the efficiency of

non-profit organizations: A critical evaluation. Managerial and Decision

Economics. 6, (1) 50-58

Odeck, J. (2001) “Comparison of data envelopment analysis and deterministic parametric

frontier approaches:An application in the Norwegian road construction sector”.

Transportation Planning & Technology. 24 (2). 111-134.

Oral M, Yolalan R. An Empirical Study on Measuring Operating Efficiency and

Profitability of Bank Branches. European Journal of Operational Research 1990;

46(3): 282-94.

Ormerod, R..J., and Kiossis, I. 1997 OR/MS publications: extension of an analysis of US

flagship journals to the UK. Operations Research, 45 179-187.

Pierskalla, W..P. 1987, Creating Growth in OR/MS. Operations Research 35, 153-156

Ray, S.C. (2002), William W. Cooper: A legend in His Own Times, Journal of

Productivity Analysis, 17, 7-12

Reisman A. On Alternative Strategies for Doing Research in the Management and Social

Sciences. IEEE Transactions on Engineering Management 1988; 35(4): 215-221.

Reisman A. (1992) Management Science Knowledge: It’s Creation, Generalization and

Consolidation. Westport CT: Quorum Books Publishing Company.

Reisman A, Kirschnick F. (1994) The Devolution of OR/MS: Implications From a

Page 23: Absolutely Positively Operations Research: 50 Years of

23

Statistical Content Analysis of Papers in Flagship Journals. Operations Research

42(4): 577-88.

Reisman A, Kirschnick F. (1995) Research Strategies Used by OR/MS Workers as Shown

by an Analysis of Papers in Flagship Journals. Operations Research 43(5): 731-40.

Reisman A, Kirschnick F. (2001) A Role for Content Analysis in Research on OR/MS

Research Literature, An invited paper. Omega; 28 247-48.

Reisman A, Kumar A, Motwani J. (1997a) Flowshop Scheduling/Sequencing Research: A

Statistical Review of the Literature 1952-1994. IEEE Transactions on Engineering

Management; 44(3): 316-29.

Reisman A, Kumar A, Motwani J, Cheng CH, (1997b) "Cellular Manufacturing: A

Statistical Review of the Literature (1965-1995). Operations Research 45(4): 508-

20.

Reisman, A., Kumar, A., Motwani, J., (2001) A Meta Review of Game Theory

publications in the flagship US-based OR/MS journals. Management Decision;

39(2): 147-155.

Romero C. (1986). A survey of generalized goal programming (1970-1982). European

Journal of Operational Research. 25, 183-191.

Romero C. (1991). Handbook of critical issues in goal programming. Pergamon Press,

Oxford.

Sarafoglou N. (1998) The most influential DEA publications: A comment on Seiford.

Journal of Productivity Analysis. 9(3): 279-81.

Schniederjans M. J. (1995), Goal programming: methodology and applications. Kluwer

Academic publishers, Norwell USA.

Seiford L. M.. (1996) Data Envelopment Analysis: The Evolution of the State of the Art

(1978-1995). Journal of Productivity Analysis 7 99-137.

Seiford, L.M. and Thrall, R.M. (1990) Recent development in DEA: the mathematical

programming approach to frontier analysis. Journal of Econometrics 46 7-38.

Sinuany-Stern Z, Mehrez A, Barboy A. (1994) Academic Departments and Efficiency Via

DEA. Computers and Operations Research 21(5):543-6.

Tamiz, M. Jones D.F. and Romero C. (1998) Goal programming for decision making: an

overview of the current state-of-the-art. European Journal of Operational Research.

Page 24: Absolutely Positively Operations Research: 50 Years of

24

111, 560-581.

Tavares , G. ([email protected])

Thanassoulis E. (2001) Introduction to the Theory and Application of Data Envelopment

Analysis: A Foundation Text with Integrated Software, Kluwer Academic Publishers,

Norwell Mass.

Thrall, Robert M., 1989. Classification transitions under expansion of inputs and outputs in

data envelopment analysis", Managerial and Decision Economics, 10 (2)159-162

Zanakis SH and Gupta SK (1985) A categorized bibliographic survey of goal

programming. Omega. 13 (2), 211-222.