Document created: 2 September 03
Air University Review, May-June 1975
Leland G. Jordan
Output measurement and management indicator systems provide information concerning the output of production processes and the relative quality or efficiency of that output. The selection of valid standards against which relative quality or efficiency can be judged is difficult but possible.
To date, the Air Force output measurement and management indicator system has not been effective in providing decision-makers with the accurate, pertinent information needed or intended. A revision to both the philosophy and methodology would appear to be indicated. Such a revision has been identified and discussed in the context of a particular management indicator system, but it applies to output measurement and management indicator systems in general. In this article the present Air Force output measurement and management indicator system is described, the reason for the lack of success of the system is suggested, some requirements for a successful system are identified, and a basic question concerning Air Force management philosophy is posed.
The stated objective of the Air Force output measurement and management indicator systems is to support and influence management by improving the capability to describe outputs and the quality availability, and utility of output measures and management indicators. Output measures are generally described as production measures, and management indicators as rates, such as cost per unit or percent of programmed production accomplished. The system, known as the Cost Center Performance Measurement System (CCPMS), which the Air Force implemented in 1972, was an attempt to provide cost-output information to managers. The governing manual, Air Force Manual 178-430, provides a statement of the CCPMS objective, describes methods for selecting and validating cost center performance measures, and discusses their use.
The CCPM system is to select an output measure, which is then related to dollar costs via regression analysis. The output measure having been selected and the regression analysis completed, the resulting equation is used to compute a standard cost. The major command can use the standard cost to estimate the expected total cost for a given level of production. The expected cost is then compared to the actual cost. However, regression analysis and statistical prediction are not exact sciences; therefore, the predicted costs are ranges, not point estimates. Figure 1 illustrates the uncertainty in such a predicted cost.
The cost-output regression provides an equation--in this example it is presented by a straight line--relating production to cost. For the line in Figure 1, if "Q" items are produced, then a cost of "b" dollars is expected. But because of the uncertainty, a cost of less than "a" dollars but more than "c" dollars is considered reasonable. That is, the regression equation enables us to say that "for a production of 'Q' units the total production cost will probably be between 'a' dollars and 'c' dollars with 'b' dollars the most likely estimate." The regression equation explains the variation in cost that results from varying the production quantity. The uncertainty, that is, "a" minus "c" dollars, accounts for the variation in cost that is not attributable to changes in the production quantity.
The standard cost for "Q", items is simply "b" divided by "Q" or the average expected cost at that point on the regression line. Clearly the standard cost is different for different production levels, since the fixed cost is spread over more or fewer units as the production quantity changes. The standard cost derived from the regression line, therefore, helps management estimate how total cost changes with production quantity.
The range a minus c" is a probability range; that is, the total cost is expected to be between "a" and "c" some specified percentage of the time. For example, "a". and "c" may be chosen so that for nine out of ten times that "Q" items are produced, the total cost will be between "a" and "c" dollars. In other words, we may be 90 percent certain that if "Q" units are produced, the total cost will be less than "a" dollars but more than "c" dollars. That certainty percentage (90 percent) or confidence level can be adjusted to, say, 67 percent or 99 percent, depending on the analyst's selection of "a" and "c."
The major command provides the relationship to the functional organization and directs that if the costs fall outside the uncertainty bounds (that is, if the total cost is less than "c" or more than "a") then a further analysis should be conducted. The idea, of course, is to provide an exception reporting system for management.
The preceding discussion has been a very brief summary of parts of the CCPM system as described in AFM 178-430. The manual itself provides a detailed description of regression analysis and the use of the resulting cost-output relationship. In fact, the manual does an outstanding job of explaining how to select an output measure and fit a regression equation. So the intent of this discussion is to provide sufficient background to allow discussion of the philosophy or theory of this Air Force output measurement system, not the statistical techniques.
CCPMS sounds like a valuable management tool. It also sounds very much like the quantitative techniques used in industry. Yet it was not accepted by the functional managers. The most frequent comment made about the system was that it did not contribute to management; that it was a wasted effort. Why? What was CCPMS missing?
The candidate answers to that question are numerous. The inclusion of costs not controllable by the manager is considered by some individuals as the primary fault of CCPMS. The extent to which the manager can control his costs is important but secondary to the major philosophical thrust of the current argument. While a major portion of the costs is military personnel costs and not within the production manager's control, several solutions to that complication are available. The essential point is that whether or not the production manager can control his costs is material only after the decision process has been cleared to function. That is, if the information is not to be used, its purity is immaterial. Decision information systems should be designed in the same way decision networks are solved--from the last decision to the first. The purity or controllability of the costs is, therefore, a secondary consideration. Product quality and the danger that managers may make short-term gains at the expense of long-run results are also secondary to the thrust of the current argument. For the purpose of this discussion, therefore, it is assumed that costs are controllable, quality is satisfactory, and no short-term suboptimizations are being made. These may be real problems, but their solution is not the intent of this discussion.
information system design
Decision information systems should be designed after the decision is identified. The information system can then be designed to collect data and present information that will assist an executive in making the identified decision. Of course, information systems are more frequently designed for a general class of decisions rather than one decision; however, the principle is that one should not build an information system and then go out, like Diogenes with his lamp, search-trig for a decision. If information systems are not useful, it may be because they do not provide information that aids in decision-making.
the efficient standard
The information presented by CCPMS is useful in estimating the dollar changes associated with proposed or planned changes in production quantity. It is not useful for improving the effectiveness or efficiency of the production operation, nor is it useful for identifying the particularly efficient or effective manager. Yet it is the efficient manager who can tell us how to improve the operation in similar organizations. CCPMS is based on regression analysis; it uses a regression equation to estimate a standard cost. But the regression equation explains the cost variation due to production-level changes and simply provides an interval estimate to account for the cost variation resulting from things other than production-level changes. The problem is that to identify outstanding managers we need to look at the cost variation resulting from management actions. That is, we need to look at the variation that the regression equation does not explain.*
*The unexplained variation includes variations resulting from managerial policy as well as other unexplained variations.
If the regression equation predicts the cost as less than "a" dollars but more than "c" dollars, we need to find the manager who consistently comes in with a total cost less than "c" dollars; he is the efficient one, the one who produces at a lower cost than expected. The analysis techniques required are known, and the data are available in the CCPMS files; therefore, it should be feasible to identify the efficient manager.
Once identified, the efficient manager becomes both a standard and a source of improved operating techniques and procedures. Again, the regression estimate is not an efficient standard; it is essentially an average, a standard of mediocrity. The hundred-yard dash is an analogous situation: the world record in the hundred-yard dash is 9.1 seconds; the average time is not recorded! Times in the hundred-yard dash are not judged as good or bad by how close they are to the average time but by how close they are to 9.1 seconds, the world record. The average time would be a mediocre standard; 9.1 seconds is an "efficient" standard. CCPMS used the average performance as a standard, but managers do not need much help in identifying the average; identifying the efficient manager is more difficult, and assisting that identification is a proper function of management systems.
That is the problem of output measurement in the Air Force so far. The final step has not been taken and an efficient standard identified. There has been a fear by functional managers that the standard costs will become a rating system. AFM 178-430 specifically denies that the standard costs will become a rating system. But it should be! Rating people--identifying the efficient and the inefficient--is a major management task and should not be denied. If Air Force managers are unwilling to take the rating responsibility, then cost-output relationships are unnecessary, for they provide decision information--useful only if managers accept their responsibility to decide.
If the decision is made to use output measurement as a tool for identifying efficient and effective managers, then other questions must be addressed. How about costs not responsive to the manager's actions? How do we judge quality? How do we assess short-term optimization? These are important questions if a management tool like CCPMS is to be utilized; they are irrelevant if the regression line continues to be the standard. The decision to extend output measurement methodology to provide for selection of an efficient rather than a mediocre standard must be reached before the secondary questions are pertinent.
It is always easy to be critical but frequently difficult to offer constructive criticism--to propose a remedy along with the faultfinding. The criticism presented here is constructive; a remedy is offered. The analysis techniques for implementing that remedy, for identifying the efficient manager, are not presented, but they do exist. Competent analysts should have little difficulty grasping those techniques. Since this article is intended for a broader audience, a detailed presentation of the analytical techniques has not been included. Suffice it to say that the efficient manager should be selected by identifying and further analyzing those bases or shops which consistently operate at a lower cost per unit than the cost-output regression says they should.
The philosophical question posed by this article concerns the desire to select an efficient standard. If the Air Force management philosophy is to satisfice, that is, to accept a merely satisfactory or average management performance, then cost-output systems are not needed. Conversely, if the Air Force management philosophy is to optimize, to manage as efficiently as possible, then cost-output systems are needed because such systems can substantially simplify the complexities involved in finding the efficient manager and hence the efficient production methods. However, effective worthwhile cost-output systems must be designed differently than the CCPM system was designed; they must be designed to locate and use an efficient standard.
Hq Strategic Air Command
Leland G. Jordan (M. S. Systems Analysis, Air Force Institute of Technology) is Chief, Cost and Economic Analysis Division, Hq Strategic Air Command. During eleven years on active duty in the Air Force, he served as a costs analyst at Hq SAC and in base-level staff positions in France, New Mexico, Thailand, and Texas. Mr. Jordan, a member of several professional associations has presented papers on models of Air Force financial management systems.
The conclusions and opinions expressed in this document are those of the author cultivated in the freedom of expression, academic environment of Air University. They do not reflect the official position of the U.S. Government, Department of Defense, the United States Air Force or the Air University.
Air & Space Power Home Page | Feedback? Email the Editor