According to Peter F. Drucker:
A manager, in the first place, sets objectives. Secondly, a manager organizes. Next a manager motivates and communicates. The fourth basic element in the work of the manager is the job of measurement. Finally, a manager develops people.
Planning, organizing, staffing, and organizational development are the key functions of top management in any organization. I believe that the top managers in the Department of Defense (DOD) have failed to perform these functions satisfactorily. Because of this failure, they have been forced to devote their attention to lower-level tasks: operating, controlling, and budgeting. Thus overwhelmed by administrative detail, they have allowed the DOD to become turgid, sluggish, and ineffective. I further argue that this failure largely reflects the incomplete implementation of organizational and budgetary reforms initiated over thirty years ago. As S. E. Huntington has correctly observed:
Criticism has been directed at many aspects of defense department organization, including, for instance, procedures for weapon system procurement. Varied as the criticisms have been, however, they have tended to focus on the strategic side of the defense establishment -- how decisions are made on overall policy, on the development of military forces, programs, and weapons, and on the use of military force. These criticisms tend to articulate in a variety of ways a single underlying theme: that there is a gap between defense organization and strategic purpose. This gap is the result of the failure to achieve the purpose of organizational reforms instituted [at the end of the Eisenhower era].
By organizational structure I mean three distinct but related things. In the first place, I mean the division of labor in the organization: dividing work into tasks or roles such as operations, logistics and transportation, and training, and recombining them into administrative units, e.g., branches, departments, bureaus, or divisions according to mission, function, and/or region. This is the organization's administrative structure -- the structure depicted in organization charts, including tables of organization and equipment. In the second place, I mean the distribution of authority and responsibility to individuals within the organization. This is the organization's responsibility structure. In the third place, I mean the organization's system of measuring and evaluating performance -- how it organizes information on inputs, costs, activities, and outputs. This is the organization's account or control structure. An account structure should be oriented to administrative units or responsibility centers -- optimally to both, since the information provided by these accounts can be used both to coordinate unit activities and to control the behavior of responsibility center managers -- or to some analytically relevant construct.
Strategy should determine structure. Strategy means the pattern of purposes and policies that defines the organization and its missions and that positions it relative to its environment. Single-mission organizations should be organized along functional lines; multimission organizations should be organized along mission lines; multimission, multifunction organizations should be organized along matrix lines. Where a matrix organization is large enough to justify an extensive division of labor, responsibility centers ought to be designated as either mission or support centers, with the latter linked to the former by a system of internal markets and prices.
Mission attributes constrain the degree of decentralization possible within an organization. Some missions require precise orchestration and coordination of complex bundles of tasks if they are to be carried out successfully. Missions that have this attribute -- for instance, performing Beethoven's Ninth Symphony or invading Europe -- must be directed by a single authority responsible for both specifying who must do what and the ultimate success of the mission. The existence of reciprocal interdependence (i.e., activities that must be carried out simultaneously to be successful) severely limits the degree to which the responsibility structure of an organization can be decentralized. Goal ambiguity -- i.e., ambiguous administrative boundaries and roles and mission responsibilities -- also constrains decentralization. Unlike reciprocal interdependence, however, goal ambiguity is usually self-inflicted.
At the other extreme, where organizational missions or activities do not directly impinge upon each other, decentralization of responsibility and authority is not at all constrained. For instance, the performance of GM's truck division does not depend on the performance of the Cadillac division or on any of GM's other operating divisions; hence GM's top management delegates full operational authority to the managers of its truck division. This same pattern characterized the authority given theater commanders in World War II. Nevertheless, as long as the missions or activities in question are performed within the bounds of a single organization, even where decentralization is complete or very nearly so, mission performance will be constrained by the total pool of resources available to the organization -- top management attention, production capacity, or space. Hence, even in a decentralized organization, it will be necessary for top management to evaluate the organization's missions and ration the limited resource among them.
The significance of common pool or common property resources is a function of the organization's time horizon. In the short run, nearly every resource is constrained. In the long run, there are no real resource constraints. Military strategy has two aspects -- raising, equipping, and organizing armed forces and using them in support of political objectives. Preparing forces for combat is governed primarily by longer-run considerations. The implication is that in peacetime, at least, there may be few if any common pool resources that have to be rationed by DOD's top management. In contrast, where combat is concerned, short-run considerations obviously predominate, since the exigencies of combat compel the use of forces in being.
Finally, there is the intermediate case in which mission performance depends upon the sequential performance of a series of tasks or functions. This means that the movement of work from one administrative unit to the next goes one-way. For example, the sale of aluminum ingots depends upon the following sequence of activities: mining, transporting the ore to a processing site, smelting and fabricating it into ingots, and finally distributing the finished product to end-users.
Sequential interdependence need not constrain decentralization, however. The relationship between sequentially dependent administrative units is essentially bilateral in nature. Where bilateral relationships are concerned, it is usually possible to set up some kind of transactional arrangement to eliminate or internalize spillovers. In most cases, these relationships can be governed satisfactorily via buyer-seller arrangements and, where they occur within the organization, by appropriate transfer prices.
Unfortunately, the failure to think these structural relationships through, to clarify administrative boundaries, functional roles, and mission responsibilities, and to develop a control structure that links each administrative unit and responsibility center to the goals of the organization as a whole is highly inimical to decentralization. This failure produces an organization that is rife with externalities and common property resources, one in which everything depends upon everything else -- in other words, an organization in which effective decentralization is impossible.
Organizations that fail to sort out the relationships between their component units -- i.e., that fail to align their administrative structure with their responsibility and control structures -- must choose between centralization and organizational chaos. Given that choice, centralization is probably the preferable alternative.
Our basic criticism of the DOD in general and the Office of the Secretary of Defense (OSD) in particular is that it has failed to clarify administrative boundaries and roles and mission responsibilities and to delegate authority accordingly. Instead, it has alternated between delegating authority to the military departments and centralizing authority in the hands of the Secretary of Defense. Each swing of this pendulum has first produced failures and excesses, in some cases arising out of the pursuit of parochial interests on the part of the military departments, and then led to renewed centralization, to a further proliferation of rules and regulations aimed at preventing error, to auditors and inspectors general charged with monitoring compliance with the rules, to overstaffing, and ultimately to incrustation, inflexibility, and immobility. Of course, I do not believe that anyone in the OSD ever consciously chose the extreme centralization that characterizes the DOD. No one ever consciously chooses obesity; it is acquired, one bite at a time, over an extended period.
Most managers within DOD would have probably agreed with Charles Hitch and Roland McKean's observation:
Decentralization of the decision-making function is an extremely attractive administrative objective -- in the military as elsewhere. The man on the spot can act quickly and flexibly. He has intimate first-hand knowledge of many factors relevant to his decision. Large hierarchical organizations, by contrast, tend to be sluggish and hidebound by rules and regulations. Much of the time and energies [of the organization's top management] are consumed in attempting to assemble at the center, the information so readily available 'on the firing line;' and since these efforts are never successful, their decisions have to be made of the basis of information both incomplete and stale. Decentralization of decision-making responsibility has the further advantage of providing training, experience, and a testing ground for junior officers. The best way to develop qualities of responsibility, ingenuity, judgment, and so on, and to identify them is to provide genuine opportunities for their exercise. Unfortunately the superficial illogicalities of decentralization are more strikingly obvious than the deadening consequences of extreme centralization.
Charles Hitch, for example, continued to stress his absolute dedication to decentralization throughout his tenure as Defense Comptroller during the McNamara era, when at the same time he was the driving force behind the installation of the most comprehensive central planning apparatus this side of Moscow.
Given the failure I have attributed to DOD's top management, one might ask how the United States ever managed to raise, equip, organize, or use its military forces successfully -- how did it manage to prevail in the last World War? The best explanation of the American victory in World War II is overwhelming material superiority. Nevertheless, as William A. Niskanen explains, prior to the end of World War II, the administrative structure of the U.S. defense establishment was for the most part aligned with its responsibility and control structures. Up to that time each of the military departments had a single primary mission and each defined its mission in functional terms: land combat was the primary mission/function of the Army; sea combat was the primary mission/function of the Navy; and tactical air support of land combat was the primary mission/function of the (Army) Air Force. Because each of the military departments had a single primary mission as well as a single primary function, it was possible, indeed often wise, for decisions about defense strategy to be made along service lines.
The rapid development of military technology and the world-wide proliferation of American commitments after World War II sundered the relationship between the armed services and their traditional functions and missions. Most of the postwar inter-service disputes have been about roles and missions, and thereby resources, which is to say organizational boundaries and proprietorship; but since property rights are lacking these disputes can only be resolved by fiat or bargaining between the interested parties, usually at someone else's expense. For example, for over forty years the Army has criticized the Air Force for its neglect of close air support (CAS) and tactical airlift. This dispute arises because the Air Force does not materially benefit from supplying air support and tactical airlift services to the Army and because it provides them at no cost to the Army. Consequently, the Air Force has no incentive to bolster its contributions to these functions and the Army has no incentive to economize on them. Several Defense Secretaries have tried to resolve this, but imposed solutions that run counter to institutional self-interest seldom last and, in this case, they haven't.
The rivalry that broke out between the military departments over the allocation of combat functions and the resources that came with functional responsibility was only the most visible organizational symptom of the changes wrought by new military technologies. Nuclear explosives created a singular, even extraordinary, new defense mission -- preparing for, coping with, and avoiding the consequences of strategic nuclear exchange. After World War II, each of the services vied for the right to deliver these awful weapons. The Air Force wanted to drop them from land-based bombers, the Navy from carrier-based bombers, and all three of the services invested resources in the development and deployment of nuclear-tipped ballistic rockets and cruise missiles. Several other technologies -- surface-to-air missiles, helicopters, etc. -- were also significant to military organization, since their use did not fit neatly within the functional boundaries of the existing military departments.
Commitments made to diverse far-flung allies had perhaps an even greater effect on the pattern of purposes that defined the postwar American military establishment and its missions than did the new technologies. These commitments required America's defense strategy to be reformulated in terms of regional defense objectives. They also required deployment of substantial conventional forces and logistical capabilities overseas, although this requirement was not fully recognized until after it became evident that the United States could not rely on the use of tactical nuclear weapons and the threat of massive retaliation to achieve its foreign-policy objectives.
In nearly every case America's regional defense plans called for operations involving a combination of ground, sea, and air forces. Not only did this subordinate combat functions to regional mission objectives, it also implied that the three military departments would have to prepare their forces for a variety of missions in different theaters, since most regional defense plans contemplated the use of units from more than one of the military departments. Depending on how their units were earmarked for various regional contingencies, each of the military departments now had several missions, the majority of which were to be achieved in collaboration with other services. This often meant that the performance of units supplied by the Air Force, for instance, often , had important spillover effects on the performance of units from the army or the navy. Not surprisingly, each of the military departments sought to internalize these externalities by assuming functional roles that had traditionally been the sole responsibility of one of the other services. Of the services, the navy carried this process the furthest, as Niskanen explains:
Each of the services developed forces in new areas as a hedge against changes in strategic concepts and military technology. An attempt was made to limit interservice competition was made in the Key West agreement of 1947, but this agreement was rapidly -- and understandably -- undermined. The Navy, whose position was threatened by the absence of a significant Soviet surface navy, was most successful in broadening its mission base. They developed an effective contribution to the strategic offensive forces, first with the carrier air forces and subsequently with the Polaris missile submarines; they also developed substantial antisubmarine forces to defend coastal waters and to protect sealift to the theaters where the large Soviet submarine force would be a threat. The Marines have been partly transformed from a navy-supported assault force to a more independent, sustained land-combat force. The carrier air forces have been reoriented to provide tactical air support for land combat in all theaters.
Considerable attention was given to defense organization during the 1950s, although much of the discussion focused on the symptoms of the problem, especially interservice rivalry, rather than the problem itself. The first solution, or at least the first that was widely supported by the defense policy community, was aimed directly at the elimination of service competition. It proposed to reestablish unity of organizational structure by creating a single uniformed service and thoroughly restructuring its component branches on strictly functional lines. This is, incidentally, how the Soviet military establishment is organized. This proposal was not adopted -- primarily because it threatened the prerogatives of the military departments and perhaps of Congress. But, even if this proposal had been politically acceptable, it made very little sense in its original form.
As Edward Luttwak has observed, military leadership and tradition are critical to the capacity of the DOD to protect and defend the United States. It is axiomatic that the performance of any organization depends upon its culture. Organizational culture is, of course, merely a shorthand way of referring to a bundle of values, traits, and norms that are shared by members of an organization, including mutual trust, personal and group loyalty, and individual attributes such as self-control or intrinsic motivation. It is what pushes people to perform adequately in difficult or highly demanding circumstances, even when it is not in their interest to do so. The traditions of the military departments and their component elements are an irreplaceable reservoir of institutional culture and values. It would be foolish indeed to discard or even threaten so valuable an organizational asset. This fact necessarily constrains reorganization and personnel options within the defense establishment and, insofar as unification threatens military tradition, it rules out that option entirely.
Perhaps even more importantly, this proposal would have reestablished the unity of organizational structure at the cost of divorcing it from organizational strategy. As we have seen, American. defense strategy was formulated in terms of missions that called for combined operations: strategic nuclear exchange and regional defense. The preparation and conduct of combined operations requires unity of command at the mission level. Making function and role central to military organization would have denied mission commanders the authority needed to orchestrate the actions of various military units supplied by different services to achieve a common objective. Instead, coordination would have depended upon the voluntary cooperation of the functional elements assigned to accomplish the mission objective, perhaps enforced by the Joint Chiefs of Staff (JCS), where they could reach consensus, or, where they could not, by the President or his deputy, the Secretary of Defense.
This prospect may have seemed less disheartening, perhaps even appealing, when Eisenhower and Marshall occupied these offices, than subsequently. But it must have seemed rather far fetched. During World War II, theater commanders were given absolute authority over the units assigned to them. When the Korean War broke out, one of the first acts of the Truman administration was to give MacArthur full command of all U.S. forces in Asia
Defense Reorganization Act of 1958
Recognizing the need for unity of operational command at the mission level, the Eisenhower administration proposed and the Congress passed the Defense Reorganization Act of 1958. This act increased the authority of the OSD and provided for the reorganization of combat command along regional lines, which could have gone a long way toward reconciling America's defense structure with its strategy. The combatant (or "unified" and "specified") commands created by the Eisenhower administration were intended to be the principal instruments of American defense policy, each using forces supplied by the military departments. Eisenhower wanted to decentralize operational authority and responsibility to mission and theater commanders, because he believed that the combatant commanders were potentially best situated and motivated to plan operations and to determine the size and composition of the forces required to accomplish their assigned combat missions as well as to carry them out. Eisenhower's commitment to decentralization was evidently based primarily upon his experience as a theater commander during World War II, but it may have been reinforced by the performance levels achieved by the Air Force under decentralization in the period following the Korean War, especially that of the Strategic Air Command under General Curtis LeMay.
The Eisenhower reforms implicitly assigned the military departments a distinctly subordinate or supporting role vis-à-vis the combatant commands. The military departments were to recruit, train, and equip forces (units) to perform combat and support roles. As I will explain shortly, if combatant commanders had also been given the freedom to select the units they wanted from among those supplied by the military departments, the Eisenhower reforms might have transformed interdepartmental rivalry from a liability to an asset. However, such a transformation would have required financial or budgetary resources to flow to the military departments through the combatant commands, and Eisenhower did not propose this system. Instead, combatant commanders must take the forces the military departments choose to supply and the military departments retain full administrative control over those forces:
the combat forces of each service are supposedly controlled by "unified" or "specified" joint service commands, but they still respond first and last to their own service fiefdoms, which train, equip, and maintain them with their own supporting commands, under the supervision of their own service chiefs, who alone decide the fate of each officer's career.
In other words, the Eisenhower reforms sought to align Americas defense structure with its strategy, but what they accomplished was to divorce completely the military's responsibility structure from its administrative structure and its system of measuring and evaluating performance.
Consequently, instead of clarifying administrative boundaries and mission responsibilities and developing a control structure that linked each responsibility center and administrative unit to a coherent set of mission objectives, the Eisenhower reforms made a vastly complicated bureaucracy even more complicated and created additional spillovers and common property resources where none had existed before. Because the cooperation of the supplying military department remained voluntary, combatant commanders found it necessary to make concessions to "buy" their cooperation using the only currency available to them. And, because it is easier for bureaucracies to add than to subtract, these concessions often took the form of long-term treaties guaranteeing the cooperating military department an enhanced role in the implementation of the combatant command's plans and activities. It is possible that these kinds of concessions may have cost some commands the capacity to achieve their assigned objectives. It is certain that they reinforced the propensity of the military departments to develop forces to perform a variety of functions and roles, even where others could supply equivalent forces or close substitutes in a more cost-effective fashion.
During the 1950s and even later, most analysts were far more concerned about the problem of allocating resources to and within the DOD than they were with its organizational structure or management controls. From here is easy to recognize their shortsightedness and to criticize them for it. But hindsight is always clearer than foresight, and in this instance their myopia was understandable. Thirty years ago, there was little evidence of a gap between defense organization and strategic purpose, let alone of its seriousness. The visible symptoms of this gap -- failure at the operational level and terminal bloat at the administrative level -- still lay in the future. At the same time, it was obvious that the proliferation of overseas commitments, rapid technological change, and the subsequent transition from massive retaliation to flexible response required urgent changes in the shape and size of the U.S. military-force structure and its logistical support capabilities.
Because the problem was defined in terms of making choices about the allocation of resources, the solution was equally obvious as well as equally shortsighted: have the OSD figure out what to do and order the military departments to do it. This solution, however, carried with it two additional problems. The first was that the OSD believed that it lacked the information needed to figure out what to do. The second was that the OSD couldn't simply tell the military departments what to do. The OSD doesn't supply resources to the military departments, Congress does. And Congress is not called upon to fund the capability to do things (i.e., perform missions, functions, or tasks). Congress provides funds to buy things. In the 1950s the military departments requested and Congress provided funds in terms of the following object of expenditure classes: research, development, testing, and evaluation; procurement; construction; personnel; and operations and maintenance. The distinction between funding by administrative component and by object of expenditure class, function, or mission is illustrated in Table 4.1.
In order to get the military departments to do what it wanted them to do, the OSD believed that first it had to figure out what the military departments needed. Then the military departments had to be told to ask Congress for funds to buy the things they needed -- to carry out the OSD's wishes. Finally, Congress had to give the military departments the funds they requested. This last was not seen as a serious hurdle, however, since congressional acquiescence could be assumed.
PPBS: or Mr. Hitch's Marvelous Budgeting Machine
To help the OSD tell the military departments what to do, a group of economists at The RAND Corporation developed a comprehensive set of accounting and budgeting procedures. Together with wholesale changes in the DOD's control structure, away from one oriented to administrative units to one oriented to functional constructs, these concepts and procedures comprised the PPBS process -- the planning, programming, and budgeting system.
The PPBS process installed in the Pentagon under Robert S. McNamara remains in effect to this day, although it has been modified in various respects (see Appendix A: PPBS). It starts with military plans, the Joint Strategic Objectives Plan (JSOP), written annually by the JCS and the service chiefs. These are then broken down by function into broad programs (e.g., strategic retaliatory forces) and are then further subdivided into hundreds of subprograms or program elements (e.g., the Midgetman system).
The OSD then issues the Defense Guidance, which steers the military departments budgeting and programming. The next step in the PPBS process is the production of the Future Year Defense Plan (FYDP). The FYDP details both the DOD's continuing programmatic commitments (its base) and its new commitments (increments or decrements) in terms of force structure (including sizes and types of forces) and readiness levels, inventories and logistical capabilities, and the development of new weapons and support systems.
Next comes programming -- in the context of PPBS programming comprises the identification of program alternatives, forecasting and evaluating the consequences of program alternatives, and deciding which program alternatives to carry out. The consequences of the DOD's programming decisions are estimated in terms of the amount, character, and timing of inputs, including all acquisitions and construction, to be funded for each program package (assuming no change in commitments) during the next six-year (originally five-year) period, although in practice only the estimates for the first two years are used by the DOD.
This process generates an array of Program Objective Memoranda. The Program Objective Memoranda are expressed in terms of current dollars and arrayed by military department, object of expenditure, and function. The first two years of the Program Objective Memoranda provide the basis for the military departments budget requests to Congress.
In the context of PPBS, budgeting means preparing an annual spending plan. The DODs budget requests set forth the DOD's policies and spending proposals in monetary terms for the next fiscal year and provides a revised estimate of the current year's spending (in the fiscal 1991 budget, the current year is fiscal l990). The DOD's spending plan is arrayed by program, but more significantly by military department, by object of expenditure, with most research and development programs, major acquisitions, and construction projects individually scheduled, by federal budget classification codes to meet the needs of the presidents Office of Management and Budget, and by appropriation categories to meet the needs of Congress.
How Capital Budgets Should Relate to Operating Budgets
Effective decentralization is possible in the private sector in part because capital budgeting and operational budgeting are treated as related but distinct processes. An organization's responsibility budget should reflect its commitments. Thus, a decision to invest resources in a new initiative should be reflected in the operating budgets of all the responsibility centers affected. The controller should revise operating budgets to reflect the increases in organizational performance that justified the decision to go ahead with the initiative. She should specify the increases in performance expected of each administrative unit or responsibility center, revise evaluative standards to take account of anticipated improvements, and assign responsibility for realizing them. But the purpose and the content of capital and operating budgets -- deciding and doing -- should be kept distinct.
The DOD makes no such distinction, however. For the most part, operating managers within the DOD may do only what their budget says they can do; they may not do what they are not authorized to do. Their budgets tell them they may incur obligations to buy things -- and that is it. Their budgets focus exclusively on resources to be acquired by individual administrative units and on the timing of those acquisitions -- on objects of expenditure rather than performance targets, on many inputs rather than a few critical outputs or results. In other words, operating managers have no real discretion to acquire or use assets; without that discretion, they cannot be held responsible for the performance of the administrative units they nominally head.
In an important book on defense budgeting, Fredrick C. Mosher advanced a proposal for a program budget that would have permitted the OSD to make essential policy decisions, but would at the same time have permitted considerable decentralization of operations. Like his counterparts at RAND, Mosher proposed program packaging to help the OSD establish strategic priorities and ration capital between major programs. He also proposed a functional account structure to force alternatives to the surface and to help the OSD identify the most cost-effective service supplier. Mosher, however, envisioned this upward-oriented program budget operating in synchronization with a results-oriented operating budget, structured along administrative lines, that would have provided the primary vehicle for internal planning and control.
In retrospect it seems unfortunate that Mosher's conception was not adopted. Not that all of Mosher's proposals were up to the mark: like his RAND counterparts, Mosher overlooked the significance of the distinction between mission and function and between responsibility and administrative structure. Moreover, his solution to the problem of reconciling the exigencies of management control with the congressional budgetary process -- lagging the operating budget behind the program budget by a year and having it start only after programs had been authorized -- was somewhat awkward. But, unlike his RAND counterparts, Mosher clearly recognized that different kinds of budgets are needed for different purposes, for which different arrangements and timing are appropriate. This insight is one of the keys to effective decentralization in any organization. During the 1950s or shortly thereafter, Congress might have willingly provided budget authority in terms of function or mission rather than object-of-expenditure categories and delegated to OSD the flexibility needed to implement a results-oriented operating budget. This was not the case during the 1970s and 1980s; it may not be the case now.
Effective decentralization of DOD would have required more than the installation of a results-oriented operating budget, however. At a minimum, responsibility budgeting and decentralization would have required prior clarification of the purpose of the major administrative units and responsibility centers in the DOD and of their relationships to each other. Thinking organizational purpose through and reorganizing to bring the DOD's structure into line with U.S. defense strategy was the singular challenge facing its top management in the postwar era. For one reason or another the OSD consistently failed to recognize the significance of this challenge or to meet it.
It can be argued that Eisenhower's and McNamara's organizational and budgeting reforms provided a necessary foundation for a rebuilt DOD -- that they failed only because their efforts did not go far enough. Building an organization is a lot like building a bridge. In both cases, one must often tear down in order to build up. In the jargon of organizational theorists, the working parts of an organization must be both differentiated -- the process of dividing work into tasks or roles -- and integrated -- recombined into administrative units. Differentiation necessarily precedes integration. Eisenhower's and McNamara's reforms effectively reduced the DOD to its basic constituent elements. Eisenhower divorced the DOD's responsibility structure from its administrative structure. And by reorienting the focus of its control structure away from administrative units to functional constructs, McNamara divorced the DOD's control structure from both. Moreover, one must often centralize in order to decentralize. Unity of command, policy direction from the top, and central control procedures are preconditions of effective intraorganizational decentralization. Eisenhower provided OSD with the authority needed to make organizational policy and strategy and McNamara used that authority to the fullest to shape the U.S. military-force structure and its logistical support capabilities to the requirements of the doctrine of flexible response.
Unfortunately neither Eisenhower nor McNamara carried his reforms through to completion. Like a bridge under construction, Eisenhower's and McNamara's organizational and budgeting reforms increased congestion and slowed movement to a standstill, but didn't shorten the trip one bit and won't until the project is finished.
This outcome cannot be blamed on ignorance. McNamara particularly appreciated the significance of organization design and development. According to Daniel Seligman:
McNamara's organizational changes began with a demand for an office in charge of organizational planning. He had set up an office for organization in his early days at Ford, and had briefly headed it himself; he was dismayed to discover that there was no such office then in existence at the Pentagon. Any such problems in this area -- e.g., about who had jurisdiction over any new management function -- were likely to be addressed by creating an ad hoc committee to look into the matter; there were scores of these committees roaming the corridors of the Pentagon during certain periods in the 1950s. McNamara announced immediately that he had to have a permanent office in charge of organizational matters. The importance he attaches to it is evidenced in the fact that he has made its head an Assistant Secretary of Defense.
McNamara also claimed that he wanted to push all decisions to the lowest appropriate level and recognized the need to make to make organizational changes to implement his management philosophy.
The most promising of the organizational design and development efforts initiated under McNamara was made by Robert N. Anthony, who succeeded Hitch as defense comptroller in September 1965. Like Mosher, Anthony recognized the significance of the distinction between deciding and doing, planning and operating. This distinction was reflected in the comprehensive resource management system he proposed, which had a separate results-oriented operational budgeting component. Moreover, Anthony saw the need for prior clarification of organizational purpose, boundaries, and relationships, and for an account structure that would tie the organization together. The account structure he proposed was firmly grounded in the principles of responsibility budgeting and accounting.
Indeed, the only significant conceptual difference between Anthony's system and standard practice in the private sector is that he proposed to establish separate funds to manage the acquisition, utilization, and disposition of fixed assets and some inventories, but this too is now standard practice in many not-for-profit organizations, including the DOD.
Anthony's system was supposed to be operational by July of 1967. But like many of the organizational design and development efforts initiated under McNamara, Anthony's proposals never really got off the ground. This failure can probably be attributed to the press of operating problems. By 1965 it was clear that nothing important could happen in OSD without McNamara's attention and support and that McNamara was preoccupied with other, more immediate concerns, first with the pacification of the Dominican Republic and then with the agonies of Viet Nam. It is hard after all to remember that you are there to drain the swamp when you are up to your ass in alligators. Besides, reorganization probably did not seem critical. To most independent observers, centralization appeared to work under McNamara. Indeed, it appeared to work brilliantly.
In his first four years as Secretary of Defense, with only a thirteen percent increase in budget authority, McNamara enhanced greatly both the survivability of Americas. strategic forces and the adequacy and mobility of its tactical forces. He tripled the number of nuclear warheads in the strategic alert forces and made massive investments in the Minuteman and Polaris missile systems. He increased the number of combat-ready army divisions from 11 to 16 and air force tactical wings from 16 to 21. He strengthened Marine Corps manpower from 175,000 to 190,000 and increased the special forces ten fold. He doubled airlift capability and quadrupled sealift, giving the United States the capability to deploy an extraordinary array of military forces at points all over the world, rapidly and efficiently. The purpose of this buildup was to increase the number of options available in a crisis, especially the capability to wage conventional war. It was so adroit that the basic force structure established under McNamara lasted through the Cold War essentially unchanged.
One irony of this is that it took thirty years for Mr. Hitch's budgeting machine to be used to help the OSD establish strategic priorities or to ration capital between major programs. The important changes in doctrine and force structure under McNamara reflected decisions made in early 1961, before the implementation of the new program format or the formal programming system. As a consequence, PPBS's predominant function throughout the Cold War was to grind out updated tables of organization and equipment and fill them with new weaponry and trained personnel.
The organizational costs of McNamara's successes were not at first evident. We know now, however, that his accomplishments were largely Pyrrhic victories. He provided the material means to project U.S. power worldwide and to respond flexibly to a variety of contingencies. But he left the DOD with a structure that was administratively bloated and organizationally incapable. Instead of crisp decision making, clear-cut control of military strategy and operations, responsibility for operational success and accountability for failure, his organizational legacy was logrolling among the JCS, a fragmented control structure, and a murky, committee-like chain of command.
From our vantagepoint, one fact stands out: the combatant commands are the principal instruments of U.S. defense policy. Flexible response requires organizational responsibilities to be divided along geographic lines. This fact is now widely acknowledged to be the key to closing the gap between defense organization and strategic purpose that appeared in the wake of the Eisenhower/McNamara reforms. One manifestation of this belief is Jacques Gansler's recommendation that the budget process be restructured to provide spending authority to the combatant commands, rather than by object of expenditure classes to the military departments. Gansler refers to this concept as mission budgeting, but he does not explain how budgeting by mission area would work -- or how it would make things better.
Abba P. Lerner provided one possible answer to these questions nearly fifty years ago. Lerner was both a socialist and an economist. As a socialist he believed that government should set goals and objectives for the economy as a whole. But, as an economist, he recognized the dysfunctions produced by comprehensive state allocation of productive assets -- land, labor, plant and equipment, and raw materials. He was especially critical of the system of detailed centralized planning and ex ante control used by Gosplan in the Soviet Union to implement its long-term policies and strategic plans. Consequently, he devoted himself to designing an institutional arrangement that would reconcile unity of command with the market and, thereby, provide a socialist alternative to state allocation of productive assets. In 1944 he published his classic work, The Economics of Control, which outlined a fully decentralized approach to socialism.
While he was inventing market socialism, Lerner observed that Gosplan's approach to economic planning and control was merely an adaptation of the Kriegwirtschaftsplan, the planning and control system used by the General Staff under Ludendorff to mobilize Germany's resources during World War I. Not surprising, Lerner concluded that if his proposal would work better than state central planning and allocation of productive assets for the economy as a whole, it should work better than the system of state allocation of productive assets in the conduct of war.
Lerner believed that the bulk of defense spending authority should be allocated directly to the combatant commanders, who would use the funds to purchase personnel and equipment from the military training and support commands and to lease logistical and support facilities and weapons systems from the private sector. According to his scheme, purchasing and leasing would be carried out by means of arm's length transactions and the whole military establishment of the U.S. would be organized as a network of markets that would operate like the private American economy does. Combatant commanders would bid for supplies of materiel and skilled personnel. Military suppliers would sell their services to the highest bidder. And, prices would be used to bring demand into line with supply.
Lerner made only one concession to centralized planning and control. He proposed to assign what we would now call the JCS responsibility for allocating the military budget to the combatant commands -- the military equivalent of final consumers. Based on their net assessment of the strategic situation, which would take account of the capabilities of potential enemies, the contribution of allied forces, the significance of the missions assigned to the combatant commands, and the relative effectiveness of their commanders, the JCS would identify the force thresholds needed to hold risks to acceptable levels. The JCS would then ration spending authority between missions so as to equate costs and risks at the margin (i.e., equate the marginal utility of the contribution made by each of the combatant commands to the overall defense strategy of the United States).
Lerner claimed several facially plausible advantages for his proposals. Combatant commanders would be able to make full use of their specialized knowledge of enemy, terrain, and weather conditions to tailor their force structure and logistical support facilities to the contingencies they face. The network of markets and arm's length transactions would supply the precise information needed by procurement and training units and private suppliers regarding the relative utility of various kinds of military units, capabilities, equipment, and skills. Finally, the budgeting and pricing procedures would maximize DOD's capability to carry out its assigned missions and responsibilities, i.e., marginal rates of substitution or tradeoffs between different military resources would tend toward equality in all missions and functions -- the necessary condition for efficient allocation of resources.
The Lerner Scheme and the New Organizational Economics
Despite the advantages Lerner claimed for his proposals, using armed forces in support of political objectives and even raising, equipping, and organizing them should probably continue to be carried out within the bounds of a single organization. My principal objection to his scheme is that it ignores the possibility of market failure and overlooks its implications for the efficiency of arm's length transactions. Consequently, I believe that Lerner overestimated the degree to which spot prices suffice to coordinate buyer-seller behavior.
Lerner also appears to have underestimated the difficulty of motivating combatant commanders and their subordinates. He evidently believed that it would be sufficient to order them to do their best and to order suppliers to maximize revenue or profit, and spot prices would automatically bring organizational capabilities and incentives into line with desired outcomes. But would the motives of combatant commanders be consistent with the national interest? Are there not important spillover effects on other commands that they would ignore? For example, should they be given funds to develop radically new equipment? According to Hitch and McKean, experience suggests that this would be a mistake. Combatant commanders have their major responsibility in the present or near future; they can be expected to heavily discount the new and untried and to avoid investments with distant payoffs, regardless of their size. They are, perhaps, even more likely to eschew shared forces, multipurpose weapons, and standardized military formations and equipment, despite the lower acquisition, supply, maintenance, and repair costs that standardization confers on the military establishment as a whole.
The U.S. defense establishment is large and complex -- it is roughly the same size and complexity as the Australian economy. Its size and complexity argues against centralized direction of operations. But the U.S. defense establishment is not -- and probably cannot be made to mimic perfectly -- a market economy, with many freely competing buyers and sellers. This is the case not only because of the need for a single set of foreign policy objectives that defines the DOD's purposes and strategies, but also because of the special interdependencies that are created in the implementation of those policies. Consider the Lerner proposals in the light of the following problems of integrating organizations and assets with military missions.
The supplier-switching problem: The efficacy of markets depends on effective product competition. Many buyers and many sellers are not absolutely necessary to make product competition effective, although their presence facilitates it. Effective product competition does require a high degree of contestability, however. A contestable market -- one with few barriers to entry or exit -- has many of the desirable qualities of a market with many buyers and sellers, even if only a single firm serves it. This is the case because contestability permits competition for the market to substitute for competition in the market.
Lerner presumed simple product competition. He implied that frequent bidding contests would be held and orders shifted among suppliers chosen simply on the basis of price. Hence, in presuming effective product competition, Lerner assumed away problems of quality, availability, reliability, and supplier loyalty. But these are potentially serious problems where the supply of combat forces is concerned, at least where services are concerned since hardware can be stockpiled. Combatant commanders cannot afford to risk disruption of supply when their need is greatest. Yet, that is precisely the risk they would run if service suppliers were free to enter or exit the market at will, selling their services to the highest bidder.
Moreover, given an adversarial approach to buying, in which an organization buys from hordes of suppliers held at arm's length, suppliers have no incentive to make long term or durable investments in plant, equipment, or know-how, since the value of those investments would be drastically reduced if the original transaction were prematurely terminated. A supplier's reticence to make investments designed to benefit a single customer often means the sacrifice of economies of scale that come with long production runs and innovations that improve the customer's performance. To use the jargon of industrial organization, an adversarial approach to buying will often cause suppliers to avoid asset specificity.
Integration backward to incorporate suppliers is the easiest way for an organization to safeguard the continuity of a valued relationship and thereby deal with problems of quality, availability, and reliability of supply and to encourage investment in specific assets. It is not the only way, however. In some cases, high levels of cooperation between buyer and seller can be achieved within the context of a long-term relationship governed by an administered contract. Examples include the long-term mutually dependent relationships that are found in the Japanese automobile industry and the partnerships that have developed over time between certain prime contractors and the military departments, e.g., Grumman and the Navy.
Second, the monopoly problem: An exclusive relationship between one of the military departments and a contractor illustrates a second problem with the kind of arm's length relationships called for by Lerner. By committing itself to dealing with a particular supplier, the buyer transforms that supplier into a monopoly. The interests of a profit-maximizing monopoly can never be precisely the same as its customer's. This is the case even where the monopoly supplier is part of the same organization as its customer. Given either a fixed budget or a per-unit price, the monopolist will always provide fewer goods or services at a higher price than would be preferred by its customer. Monopolies are very difficult to manage at arm's length. Theory tells us that they can be managed by means of cost-based, two-part tariff mechanisms -- e.g., the incentive-price mechanisms that have been developed to govern the behavior of prime contractors in the development and production of major weapons systems. But two-part tariffs must be very carefully calibrated to produce an efficient solution. Moreover, careful monitoring of accounts and, in some cases, preaudit of expenses are needed to offset the seller's incentive to cheat or to misreport.
Given sufficiently extensive external economies of scale and scope, asset specificities and indivisibilities, or the like, it is often easier to internalize the relationship than to try to manage it at arm's length. For example, unbalanced transfer prices can be used to manage the relationship between customer and monopoly supplier within an organization. They are easy to administer to produce outcomes that are satisfactory for the organization as whole.
Third, the asset accounting problem: Institutional design is largely concerned with allocating property rights and accounting for the ownership and use of assets. Its goal is to provide incentives to efficiency. For example, the DOD relies, in part, on quasi-vertical integration to avoid the asset specificity problem: it retains ownership of research and development and much of the fixed plant and equipment that defense firms use to produce military hardware. Where armed forces are concerned, the accounting problem is doubly difficult because efficient use of military force, personnel, and material, means one thing in peace and quite another in war. It is hard to imagine how ownership and use of these assets could be transferred from one independent organization to the next, as called for by Lerner, without biasing outcomes. Would a leasing agency permit its assets to be taken into combat? Or, if losses in combat were charged to the combatant command's account would not that deter aggressive action on the part of the commander? The incentive here would seem to be the reverse of that intended by the British Admiralty when they executed Lord John Byng. Again this is a problem more easily handled within an organization than between organizations, if only because it is much clearer to everyone concerned that the relevant accounts are merely numbers that can be moved around to produce the behaviors desired.
These problems are not dispositive, however. The new economics of organization makes us aware of a variety of issues that are relevant to the boundary between the market and the organization: transactions costs, principal-agent conflict, external economies of scale and scope, asset specificities and indivisibilities, contestability, and the like. But it has not provided us with absolute guides to the location of the boundary. It is possible -- although not, in our opinion, likely -- that some future Lerner will explain how they would organize arm's length transactions to accommodate the functional interdependencies -- reciprocal, sequential, and pooled -- that pervade America's defense strategy.
I have given a great deal of space to Lerners ideas because they go to fundamental principles of military organization and institutional design and also because they point in the right direction. What Lerner proposed for the U.S. military establishment was decentralization -- total, complete, and without qualification. I believe that decentralizing the DOD makes good sense. An organization should be as decentralized as possible. But our starting point is the DOD's existing, extremely centralized regime. It is probably not possible to take decentralization of the U.S. military establishment as far Lerner would have. Nor would it be desirable to do so. Even if his proposal had no obvious analytical flaws, its extreme radicalism is unsettling.
Decentralizing the DOD
In contrast to the difficulties and uncertainties that surround the Lerner proposal the conceptual foundations of intraorganizational decentralization and their application to the DOD are fairly well understood.
Align Strategy with Structure. An appreciation of the distinction between mission centers and support centers is central to a resolution of the DOD's strategy-structure mismatch. Mission centers contribute directly to the organization's objectives. They are the principal instruments of organizational policy. Support centers supply goods or services to other responsibility centers in the organization -- i.e., they perform functions for other support or mission centers.
According to contemporary military doctrine and strategy, the combatant commands are the principal instruments of U.S. defense policy. Contemporary doctrine and strategy assigns supporting roles to the military departments. They are supposed to supply combat and support units to the combatant commands. The units they supply to the combatant commands are organized along functional lines, but the combatant commands are not. In other words, the DOD's mission structure does not follow functional lines. The DOD is a multimission, multifunction organization.
This fact now seems self-evident. For some reason, however, earlier commentators failed to see that the Eisenhower reforms in combination with the doctrine of flexible response adopted during the McNamara era made the combatant commands the DOD's mission centers and assigned to the military departments the role of support centers.
Decentralize where Possible. Fortunately, performance of the peacetime support role assigned to the military departments depends upon the sequential performance of a series of tasks, as does the provision of these units to the combatant commands. The military departments are supposed to recruit, train, and equip combat and support units and to supply those units to the combatant commands. Thus the relationships between the responsibility centers and administrative units that comprise the DOD are essentially bilateral in nature. As I have noted, where bilateral relationships are concerned, it is usually possible to set up some kind of transactional arrangement to eliminate or internalize spillovers. In most cases, these relationships can be governed satisfactorily via buyer-seller arrangements and, where they occur within the organization, by appropriate transfer prices.
If these claims are accepted, then aligning the DOD's organizational structure with its strategy should be fairly straightforward, even easy: make the DOD's basic organizational structure a matrix, with the Joint Chiefs of Staff at the top, the combatant commands or mission centers the rows, and the military departments or support centers the columns; issue budget guidance to the combatant commands and permit them to buy forces from the military departments, as shown in Figure 4.1.
Like Lerner, I want the combatant commands to be able to take advantage of competition whenever possible. However, I do not want them developing military technology or organizing and equipping the military units assigned to them. Paradoxically, the situation that led to centralization under McNamara -- competition among the military departments in the functional areas -- is the chief prerequisite for decentralization to the combatant commands. Many of America's existing weapons systems and logistical concepts are the products of interdepartmental competition.
Under decentralization, each of the military departments would continue to be responsible for performing a variety of specialized functions, some partly complementary and some partly substitutable for those provided by the other departments. However, where the status quo discourages the development of capabilities outside of the military departments' assigned functional areas, I would encourage each department to develop new capabilities, weapons systems, formations, and tactical concepts designed to compete with those provided by the other military departments. As Martin J. Bailey has explained:
An important factor that tends to mitigate difficulties in monitoring the performance of the military services is that many of the activities and systems of one service are at least partly substitutable for those of another. Tactical air forces are operated by the Air Force, the Navy, and the Marine Corps, and even the Army; and in some tasks these forces serve the same objective as does artillery. Deployment capability can be provided by air transport, operated by the Air Force, by sea transport, operated by the Navy, and by forward stockpiling in the theater of potential use, operated by the service whose equipment is stockpiled. Because of the competition among substitutable systems, any temptation for a service to try to enlarge its budget by carrying performance requirements to expensive excess or by understating the capabilities of its systems will be tempered by the prospect that a reduction in its budget might result instead and that its systems might be partly replaced by those of another service.
J. A. Stockfisch also argues that this budgetary approach would provide a way to harness interdepartmental rivalry to serve a useful purpose. He reminds us that competition between the military departments to perform specialized functions will serve useful ends only if a higher authority clearly resolves mission requirements and objectives and if the combatant command has a rational basis upon which to make choices:
Operational testing and objective evaluation procedures to treat combat readiness are critical if not necessary aids to perform this function. Thus if the Marine Corps shows that it can produce combat units of equal proficiency to the Army, it should get some of the Armys budget.
I interpret this to mean that the combatant commands must have satisfactory information on unit costs, as well as performance, if they are to choose wisely between forces supplied by the military departments. While it seems likely that, under decentralization, the JCS and the combatant commands would want to maintain their own organizations for testing and evaluation, the real key to providing the information they need is the development of a sound cost-accounting system within the DOD.
Establish an Account Structure that is Firmly Grounded in the Principles of Responsibility Budgeting and Accounting. The DOD must impose common accounting standards and pricing policies on the military departments. Again, fortunately, Robert N. Anthony showed us how this would work almost thirty years ago.