OR/MS Today - February 1997 - Molifying Management

February 1997 € Volume 24 € Number 1

The Trouble With Optimal

Better management from better analysis:
Educators need to rethink why OR/MS belongs in
business school curricula

Be not the first by whom the new are tried,
Nor yet the last to lay the old aside.
-- Alexander Pope, "An Essay on Criticism" (1711)

By Francis J. Clauss

The analytical tools we call operations research or management science have a proven record of success. With them, corporations and consumers have saved billions of dollars every year. We've shown municipalities how to provide better police protection and health care. We've shown armies how to win wars. The list is endless and dazzling.

So how come teaching OR/MS has become such a hard sell, a near orphan in the business schools which fathered it? Why do nationwide surveys show that OR/MS is fighting for its survival in one business school after another -- and in some has already lost the battle? Why have job prospects become bleak, with a 20 percent drop in the employment of OR/MS workers between 1990 and 1992, as Doug Samuelson pointed out recently in OR/MS Today [1]? Have we become victims of our own success?

The news from the real world is that the customer is the new "king of the hill." One corporation after another is revising its strategies for survival by listening to buyers and following their lead. We had better listen too. Peter Bell, president of the International Federation of Operational Research Societies, put the need very succinctly a year ago: "Those of us teaching in business schools are really the marketing front line for management science and operations research. If we can't persuade our business graduates that MS/OR has value to corporations, then the battle is over, and MS/OR is done." [2]

The good news is that we are starting to heed the market demands for our teaching, albeit slowly. New OR/MS texts are abandoning algorithms -- a painful surrender for those who appreciate the elegance of Professor Dantzig's Simplex Method and the Reverend Bayes' Theorem. Spreadsheets are displacing special software. They provide more than enough power for undergraduate classes and for MBA students who are not quantitative majors [3]. With spreadsheets, students and managers can formulate problems in a language with which they're comfortable. They no longer need to translate solutions obtained in algebraic notations back into plain English so that their bosses can understand and be convinced.

Yet the questions still nag. Are we still teaching OR/MS as arithmetic gamesmanship that might be better taught in schools of mathematics? Is the shift to spreadsheets a real change in our teaching philosophy, or is it simply the use of new tools to solve old problems? Does solving more examples equate to teaching better management? Are we instructing students what to do with solutions after obtaining them? Are we truly aware that management science is two words, and that our customers place the emphasis on management rather than on science?

Optimal Solutions May Not Be Good Management
Optimization techniques are a foundation of OR/MS. Solutions that are optimal in the theoretical sense, however, can be bad management and disastrous for those who accept them blindly. OR/MS teachers and texts must motivate students to examine the consequences of implementing bad "optimal" solutions and recognize when, as future managers, they should intervene and change conditions that are harmful. We should not mislead students into believing "optimal" is a synonym for "good" or "excellent."

Multiperiod Planning: To illustrate this point, consider how well we address the management side of multiperiod production planning. Optimizing such a plan is a typical example in many old texts and a few new ones that have adopted spreadsheets. A typical optimal solution determines the number of units to be made during successive periods on regular time and overtime (and sometimes subcontracting as well). The optimal solution is the plan with the minimum total cost.

In the solution to this type of example in one recently published text, almost one-third of a firm's capacity was idle over the six-month planning period. Demand, moreover, was dropping during the last two months. Clearly, intervention was needed to do something promptly to head off the costly effects of these conditions. Yet the six-month plan was described as optimal, and the large imbalance between capacity and demand was overlooked. Readers were left with the understanding that because the plan was optimal, it represented good management and no immediate corrective action was necessary.

Having demonstrated how to obtain the optimal solution, should a teacher or text simply move on to the next problem? To do so is to follow what a mathematics professor once sanctioned after demonstrating to students how to solve a knotty problem: "Now that we know how to obtain a solution, we immediately lose all interest in it. Next problem, please." However acceptable such a teaching philosophy might be in a math department, it's hardly a proper way to teach management. Where mathematics ends is where management begins.

The purpose of a multiperiod production plan is not, after all, simply to identify how best to operate under a given set of conditions. Plans are an opportunity to examine the given conditions and recognize whether they need to be changed. It's troubling to find texts that walk away from so-called optimal solutions that leave much of a firm's capacity unused or, on the other hand, that require excessive amounts of overtime and subcontracting. Implementing either solution will balloon costs and make it difficult to sell at prices that are competitive and cover production costs. That's NOT good management. Managers, after all, are paid to recognize when changes should be made -- and then to make whatever changes are appropriate.

OR/MS teachers need to teach students to examine plans and recognize where corrective actions need to be taken. As managers, they may need to reduce the size of their work forces, hire additional workers, buy more equipment, enlarge the facility, or subcontract excess capacity that would otherwise be idle. As managers in one function, such as production or operations, they will need to work with their counterparts in others, such as marketing and finance, to provide adequate resources for satisfying customer demands in a timely and cost-effective manner. Knowing what to do with an answer can be more important than knowing how to obtain one. If you don't know, why bother?

Staffing: Optimal solutions to staffing problems typically identify the minimum total number of workers required to satisfy requirements on specific days or shifts. Having provided that and little more, many texts simply go on to the next problem. We have seen optimal plans in recent texts with as much as 50 percent overstaffing on a given day and no more than the minimum required on most other days of the week -- and no recognition that these conditions are anything but the best.

A well-taught manager would examine such a solution before accepting it. If there is excessive overstaffing on any one day or shift, excess personnel might be transferred to other days or shifts to obtain a better balance without increasing the number of workers above the minimum identified in the optimal solution. Excess staffing on unpopular shifts or days, such as weekends, might be shifted to weekdays. This leads to using a second criterion for selecting an optimal solution that also satisfies the first criterion. Aside from doing a better job of teaching management, solving the complete staffing problem can be an effective introduction to some of the more complex problems of multiobjective optimization that generally are treated in later chapters of OR/MS texts.

Better than Optimal
OR/MS texts should not leave students with the impression that a manager's job is finished once a satisfactory optimal solution is obtained. Even a good optimal solution is seldom a final answer to what a manager should do. Competition continually drives managers to do better. Case studies can be structured to show how successful managers are continually challenging optimal solutions and seeking still better ways to perform their functions and responsibilities. Scenarios can lead students through a succession of stepwise improvements to maximize profit or minimize costs. They can demonstrate what works and what doesn't.

Communicating Results
Many of us who have worked in industry recognize the truth of the old saw: "A manager would rather live with a problem he cannot solve than accept a solution he cannot understand." Quantitative analyses fail if technicians who generate solutions cannot communicate their results to other members of a company. As OR/MS Today editor Peter Horner pointed out nearly two years ago: "The days when a corporate math whiz could disappear into a corner cubicle, crunch numbers on his own for a month, and emerge with a convoluted solution that nobody else could understand, let alone implement, are numbered. Now you have to know how to work in teams, make presentations, sell yourself and your ideas, and above all, communicate. Communication is king." [4]

An OR/MS text should encourage students to prepare effective management-quality presentations that justify any recommended plans for action that follow from their analyses. Texts should take advantage of the ability of spreadsheets to prepare tabular outputs that clarify the consequences of implementing solutions and that simplify comparisons between alternate courses of action. They should also provide examples of charts and graphs that recognize "a picture is worth a thousand words."

Leadership and Teamwork
Business Week's 1996 report on "The Best B Schools" [5] cites the following comment by a job recruiter for a major U.S. corporation: "Ten years ago, these schools made little if any effort teaching leadership or teamwork. Today, they're doing a much better job in these areas."

Teaching teamwork is a special challenge to faculties. Most teachers, like most students, specialize in a particular functional area, such as accounting, finance, marketing, personnel, taxation, or production and operations. Yet as educators, we justify OR/MS as a cross-functional discipline that cuts across the divisions of corporate management. Should not our teaching and our textbooks reflect this view?

Apart from its importance as an OR/MS subject, project management offers an opportunity to demonstrate to students the leadership role of managers in organizing, implementing, directing and controlling teamwork across the lines of functional organizations. To do this, the chapter on project management in an OR/MS text must do more than teach critical path scheduling and identify which activities to crash in order to expedite a project. It should balance its treatment of quantitative matters with attention to:

the relationship of a project manager to a company's functional managers,
the development of a work breakdown structure (WBS) that responds to the project's needs,
the development of work plans that follow the WBS and provide the estimates of activity times that are the basis for critical path scheduling,
the relationship of the WBS to a firm's cost accounting structure and
the information loop that's vital to monitoring and controlling a project's execution.

Critical path methods are not the essence of project management. Teamwork is. Are we teaching project management as a means of promoting teamwork across corporate boundaries?

Choosing and Validating Models
Students have difficulty choosing appropriate models and verifying their correctness. Forecasting offers special opportunities to teach students how to do this. The data used to develop a statistical model of the past can also be used to validate the model. An OR/MS text should show how to test a model's validity, how to rule out invalid models, and how to select the most appropriate model from the choices available. Students should be taught not to expect success with their first choice -- that they may need to try several models before finding one that's valid. They should be shown how to examine models critically, because the results of their forecasts will become the basis for many of their management actions.

Having developed a valid forecasting model, a model builder needs to show how accurate his or her forecasts are. Forecasts, after all, are never precise. Students should be taught that a forecast is not a single value but a probability distribution about an expected value.

Standard forecast errors and confidence limits are important concepts that influence operating tactics (e.g., setting safety stocks for inventory management) and evaluating the risks of capital investments. Managers earn their salaries by knowing how to profit from consumer demands that are higher than forecast, and how to minimize downside risks when they're less.

Good forecasts are usually a combination of good statistics and good judgment. Teachers and texts should discuss how various types of judgmental models are used to adjust statistical projections of past trends to correct for changes in economic, social, and political forces.

An OR/MS text should also show how managers can recognize when forecasting models need revision in order to continue to provide useful estimates of future customer demands. Students should be taught the various types of tracking signals and control charts, similar to those used in statistical quality control, for monitoring actual values against forecast values. Such tools detect when a forecasting process is out of control and needs revision. Quality control should be as much a concern for forecasting as for any other process.

Forecasting is not an isolated business activity. It's a key element of the forecast-plan-action-control loops used in management information systems. We need to point this out to our students and show them how to use forecasts are management tools.

Inventory Management
Chapters on inventory management should be more than collections of EOQ models. Much of the success of Japanese industry was in recognizing the implications of the models for reducing setup and holding costs. Their success led to the concepts of better ordering policies and just-in-time (JIT) and lean management philosophies that have revolutionized management worldwide. Having participated at the birth of this revolution, should OR/MS now delete it from its teachings? Should not discussions of JIT and lean management be a feature of OR/MS classrooms and texts?

Should we ignore management tools like ABC inventory analysis, which direct managers' attention to where the best chances for substantial payoffs may be hiding? Apart from its usefulness as a management "discovery" tool, ABC inventory analysis is an opportunity to demonstrate the interaction of quantitative analysis with a company's databases, a linkage of increasing importance. Is ABC analysis so simple a tool that it's beneath the dignity of our profession and textbooks?

Quality Control
Quality has become a primary strategy of corporate management. Among the most important responsibilities of today's managers are deciding the level of quality to provide in products and creating information systems that monitor performance and ensure satisfying customers' expectations for quality. If we condemn corporate managers for being slow to realize quality's importance and reluctant to respond until forced by competition, should we not have the same feelings about its omission in the OR/MS texts we use to teach management to our students?

Process control charts are statistical techniques (that's the quantitative side) for controlling quality. Industry is embracing them and hanging them on shop walls. Good managers use them as psychological tools for providing feedback and motivating workers to do better. Why do OR/MS texts largely ignore them?

Concluding Remarks
In all fairness, the shortcomings that OR/MS Today and our other professional journals bemoan are understandable. In years past, when computer software required students to formulate problems in algebraic notation, with decision variables represented by such abstract terms as Xl and X2, it was natural for us to emphasize an algorithmic approach. And with that emphasis, we perhaps lost our focus on what OR/MS should be. We found it easy to teach mathematics. A generation later, we're still finding it difficult to teach management.

Spreadsheets can help reverse this trend. They simplify the use of quantitative analysis. Students no longer need fear the dreadful number-crunching dragons that lay hidden in algorithmic mazes. They can formulate problems in terms they can understand, using tools with which they feel comfortable.

But spreadsheets provide much more than simply a simpler tool to solve problems. With less need to focus on algorithms, we have a wonderful opportunity to move on to teaching OR/MS as a tool for better management.

As business educators, we need to rethink why OR/MS belongs in school curricula. We need to re-examine our teaching philosophies and goals -- not just our teaching methods. The symptoms of our troubles lurk in our literature. We need to ask more of our colleagues and our textbooks than we have in the past. We need to break from traditions that are leading to a declining role for OR/MS in both academia and the corporate world. Today's managers need to know how to manage as well as crunch numbers, and how to use solutions as well as how to obtain them. We must challenge students to question whether optimal solutions are even good solutions. As managers, they will need to change conditions for which a bad solution is optimal. For the same reason, we need to change too. We need to teach good management as well as good mathematics.

Change is difficult, and teachers whose classroom performance is rigidly tied to the past will no doubt remain there -- much to the detriment of both their students and the OR/MS profession.

References

Samuelson, Douglas A., "Promising Future, Disappointing Present," OR/MS Today, April 1996, pp. 42-45.
Horner, Peter R., "Money Talks," OR/MS Today, October 1995, pp. 30-33.
Winston, Wayne L., "Software Decisions," OR/MS Today,October 1996.
Horner, Peter R., "Shop Talk," OR/MS Today, April 1995, pp. 20-24.
Special Report, "The Best B Schools," Business Week, October 21, 1996, pp. 110-122

Francis J. Clauss, Ph.D., is a senior adjunct professor of Business at Golden Gate University, San Francisco, Calif.

For more information, put the number 3 in the appropriate space on the
Reader Service Formbr>

E-mail to the Editorial Department of OR/MS Today: orms@lionhrtpub.com

Lionheart Publishing, Inc.
2555 Cumberland Parkway, Suite 299, Atlanta, GA 30339 USA
Phone: 770-431-0867 | Fax: 770-432-6969
E-mail: lpi@lionhrtpub.com