Anthony Downs
[This is a chapter from Inside Bureacracy by Anthony Downs, a RAND Corporation Research Book published in 1967.]
Under certain conditions, extensive growth in a large bureau leads to a pattern of change we shall call the rigidity cycle or the ossification syndrome. This cycle does not occur in all bureaus; but it is sufficiently important for us to describe its eight phases in detail and then discuss its overall implications, using concepts developed in the two preceding chapters.^[Many of the specific ideas in this chapter were suggested by Robert Perry of The RAND Corporation, whose stimulating examples and discussions also generated the basic concept described herein.]
The first phase of the rigidity cycle occurs when an operating bureau (or set of bureaus) greatly expands. The larger it gets, and the faster it grows, the more likely it is that the entire cycle will occur, though total size is a more powerful cause than speed of growth. As the bureau grows, its top officials suffer from an increasing leakage of authority. Their efforts to counteract such leakage constitute the second phase of the cycle. This in turn leads to the third phase: a growing rigidity of behavior and structure within the bureau. In the following paragraphs, we will discuss the specific causes of such increased rigidity.
One of the devices normally used by leaders of the operating bureau to improve their control is a separate monitoring agency. This tends to increase the rigidity of the operating bureau in three ways. First, the monitor imposes ever more complex and ever more restrictive regulations upon the operating bureau. As a result, the latter may have itself in a virtual straitjacket of rules hardly conducive to flexible behavior. Second, the operating bureau must consume more and more of its resources to satisfy the increasing demands of the monitors for information and written reports. Third, the bureau also tends to devote ever more resources to figuring out ways of evading or counteracting the monitors' additional regulations. The last two effects reduce the proportion of the operating bureau's total resources devoted directly to pursuing the goals of its top-level officials.
As the bureau grows, its internal division of labor tends to embody ever more intensive specialization. This creates more difficult coordination problems because more people are involved in each decision. Also, there is often a loss of overall perspective because each task is fragmented into tiny parts. Hence individual specialists think less about making the whole operation work well, and more about increasing the sophistication of their own fragment. Finally, greater specialization leads the monitoring agency to demand approval rights over all major decisions as a means of coordinating "spillovers."
As a result of these three factors, it takes longer to get each important decision made. The bureau's operations stretch out over time, and their flexibility declines. True, intensive specialization produces important economies of scale which improve the quality of the product, and may even cause faster performance of each step in making it. But beyond a certain degree of specialization, greater speed in performing each operation is nullified by greater sophistication of each step and more time spent coordinating the increased number of steps.
The greater the hierarchical distance between low-level officials and the points where final approval of their decisions can be obtained, the more difficult and time-consuming it is for them to carry out their functions. Yet when a bureau expands as described above, key decision points tend to escalate to higher and higher levels of the hierarchy.
This occurs for three reasons. First, since each specialist deals with only a small fraction of the entire project, authority rises to a high enough level so that the efforts of many specialists can be coordinated by one superior.
Second, top officials constantly search for more effective means of control. Therefore, if officials in one section of the bureau develop a particularly effective form of organization, their superiors often attempt to impose the same organization on other sections. However, this has three effects that increase the rigidity of the bureau's operations.
-
A managerial form well suited for one section is almost invariably not as well suited for others.
-
When the managerial form is used by several different bureau sections instead of one, overall coordination must be shifted upward to a higher level.
-
When a higher-level official begins administrating this form in several bureau sections, he normally adapts it at least in part to the needs of sections other than the one in which it was invented. This weakens its effectiveness in that section.
The third cause of decision-point escalation is the promotion of zealots who are strong supporters of certain managerial techniques they have found effective in lower-level positions. They often impose these techniques upon all the bureau sections that were formerly operated by their peers but are now under their own authority. This has the three rigidity-increasing effects described above.
As a result of the above trends, the increasing size of the bureau leads to a gradual ossification of operations. Low-level officials find that almost all decisions are delayed for long periods, and merely getting a decision made requires a great deal of persistent effort. Since each proposed action must receive multiple approvals, the probability of its being rejected is quite high. Officials must devote immense efforts to filling out reports, carrying out operationally superfluous procedures, preparing elaborate justifications for past or potential actions, and hosting numerous inspectors. In addition, it becomes extremely difficult to get novel procedures approved. If a proposed action is radically new, it does not fit control processes developed from past experience. Moreover, the need for multiple approval means that one conservative high-ranking official can seriously delay novel procedures by calling for more studies.
Thus the bureau becomes a gigantic machine that slowly and inflexibly grinds along in the direction in which it was initially aimed. It still produces outputs, perhaps in truly impressive quantity and quality. But the speed and flexibility of its operations steadily diminish.
At this point, the bureau or one of its parts is suddenly instructed by its sovereign to undertake an urgent task. The task is complex, calls for novel research or operations, requires many resources, and must be accomplished as fast as possible. Examples are building the Polaris and Atlas missile systems. Based upon their experience with the bureau, top level officials realize that it is incapable of carrying out this task. Its cumbersome machinery cannot produce results fast enough, and its anti-novelty bias may block the necessary innovation.
Therefore, a new organization is set up for this task outside the normal operations of the bureau. It is much smaller than the bureau as a whole, though it may contain many members. These members have somewhat broader capabilities and are more competent than the bureau's average members, since they have been specially picked for this task. They are also exempt from normal rotation "for the duration," so turnover is low. The new organization is not integrated with the bureau's hierarchy, but reports directly to its top-level officials. Moreover, it contains enough specialists and has enough resources so that it is not dependent upon the bureau's regular chain of command for major services. It is exempt from almost all existing controls, regulations, and procedures, and is free to invent its own. Finally, it has high priority access to resources, so its allocation requests need not compete directly with all other possible users of resources within the regular bureau.
Such a special organization clearly enjoys privileges that could not be extended to the entire bureau. It would be impossible for the whole bureau to have smarter than average members, small size, and top priority access to resources. Hence the characteristics of this type of special organization cannot be considered a blueprint for reforming the entire bureau.
However, these special privileges allow the separate organization to tackle its specific task with unusual ability, imagination, decisiveness, speed, and low costs of coordination and control. Moreover, its members develop extraordinarily high morale and zeal because of their unusual freedom, the obvious significance of their task, and their ability to develop close working relationships because of the bureau's small size and the absence of constant rotation. As a result, the organization's productivity is extremely high, and its "break-out" from the regular bureaucracy usually quite effective in accomplishing the job--at least initially.
Experience shows that special organizations of the type described above rarely maintain their high productivity for more than a few years, if that long.^[Kelly Johnson's special "Skunk Works" at Lockheed Aircraft has managed to maintain high productivity over a long time period by continuing to operate with a great many special privileges, such as visitor restrictions, reduced reporting controls, and extreme secrecy. I am indebted to Thomas K. Glennan, Jr. of The RAND Corporation for pointing out many of the special features of this uniquely productive organization.] Their unusual concentration of powers and talents gradually dissipates for four major reasons.
First, if the organization is successful in carrying out its special task, it generates requirements for interaction with portions of its own bureau, or other bureaus. Once the special organization begins to deal with more ossified bureaus, it must yield some of its decisionmaking power to them. Thus, the success of the special organization often destroys its isolation--and therefore its advantages.
Second, although the special organization is initially designed to be nearly autonomous, there is almost always some vital function that cannot be completely removed from its normal bureaucratic setting. An example is obtaining final approval for use of funds (as opposed to final approval for technical decisions). This function continues to operate under relatively ossified decisionmaking procedures, becoming more and more of a bottleneck as time passes. Thus, the impossibility of initially creating complete autonomy for this organization at middle and lower levels of the hierarchy causes an ever-increasing debilitation of its special capabilities.
Third, events occur outside the purview of the special organization that are directly relevant to its assignment. For example, while the Air Force ballistic missile program was concerned with Thor and Atlas, the Russians launched Sputnik I, the Navy invented the Polaris concept, and the National Aeronautics and Space Administration was created. The Air Force missile program was caught up in relationships with other bureaus not subject to its quick-decision procedures. Such unforeseeable relationships sometimes alter the whole definition of the special organization's assignment, and subordinate some of its key decisions to procedures outside its control. Thus, the dynamic environment of its task also prevents it from isolating itself for very long.
Fourth, as time passes, other issues emerge that are just as critical--or more so--than the task assigned to the special organization. As a result, the bureau's top-level officials begin focusing their attention on other matters. But its privileged status is essentially based upon having top priority in their decisionmaking, as well as top priority access to resources. Thus, the passage of time weakens the ability of the special organization to retain the concentrated interest of the bureau's topmost officials, which in turn undermines its special privileges.
As a result of these factors, the high-productivity phase in the special organization's life gradually comes to an end. It loses its "special" nature and merely becomes another section of the bureau struggling under the normal weight of rules, regulations, and agonizingly slow decisionmaking procedures.
Based upon the continuous growth of the British Admiralty and Colonial Office during periods when the navy and the British Empire were shrinking, C. Northcote Parkinson concluded there was no relationship whatever between the number of persons employed in administration and the size of the task to be administered.^[C. Northcote Parkinson, Parkinson's Law and Other Studies in Administration, pp 2-13.] A similar nonrelationship exists between the number of persons monitoring bureau functions and the magnitude of those functions. Consequently, shrinkage of a bureau's basic activities usually does not lead to a proportional decline in either the number of monitoring personnel or the controls they impose upon operating sections. This disproportionate survival of monitoring agencies occurs for three main reasons. First, once a monitoring agency has assembled the minimum necessary staff, genuine economies of scale may enable it to greatly expand output without adding many more staff members. Conversely, when it contracts output drastically, it may still need to retain that minimum-sized group. Second, most top-level officials seek to increase control over their subordinates rather than to reduce it. Hence they may deliberately retain the same sized staff to monitor a smaller operating section, knowing this will intensify control. Third, members of the monitoring agency are sometimes organizationally closer to the bureau's topmost officials than members of the operating bureau. Hence topmost officials may retain large control staffs in order to maintain their own power, income, and prestige, or those of their bureau.
If a bureau retains a large monitoring staff to control a drastically reduced operating staff, the former can impose much greater burdens of regulation upon the latter. Hence, when a bureau's basic social functions gradually decline in importance, its remaining operations often become encased in ever thicker and less flexible layers of regulations. This result of a fundamentally declining growth rate is fully consistent with our analysis in Chapter II of how shrinking size causes a bureau to become more conserver dominated.
The preceding analysis should not be construed as a condemnation of all bureaus. Unlike some of the severest critics of bureaucracy, we do not contend that most bureaus are so ossified they should be abolished.^[Ibid; and The Law and the Profits, passim; G. Tullock, The Politics of Bureaucracy, pp. 221-224, and Michel Crozier, The Bureaucratic Phenomenon (Chicago; University of Chicago Press, 1964), pp. 175-208.] Observers who claim that major portions of any nation's bureaucracy are beset by rigor mortis usually overlook the pressures exerted upon most bureaus by their power settings. If a major bureau becomes absolutely rigid in its behavior, its sovereign will soon begin hearing loud feedbacks from clients, suppliers, regulatees, rivals, and allies. Hence the bureau will find itself under strong pressure to become more flexible. The rigidity cycle is least likely to occur in bureaus that are under strong and constant pressure from such feedbacks. Conversely, it is most likely to occur in bureaus that are insulated from feedbacks. This implies that bureaus in democratic societies are less likely to ossify than those in totalitarian societies. It also implies that, within a democratic society, bureaus that serve the electorate directly are less likely to ossify than those that do not. Excessive rigidity in such bureaus as the State Department, AID, and the military services, therefore, may persist for extensive periods.
The rigidity cycle is much more likely to appear in communist countries than in most western nations for two reasons. First, most bureaus in non-democratic societies receive weaker feedbacks. Second, the bureaucracies in at least two communist nations--China and Russia--are vastly larger in absolute size. Faced by enormous hierarchies with dozens of levels, top-ranking officials are compelled to establish giant monitoring bureaus that develop complex hierarchies of their own (such as the Communist party).
In spite of their control efforts, sizable segments of the bureaucracy in these large systems are likely to operate under what Tullock refers to as bureaucratic free enterprise.^[G. Tullock, The Politics of Bureaucracy, pp. 167-170.] This denotes almost completely autonomous activity uncontrolled by any top-level, centralized agents. Bureaucracies in communist nations are undoubtedly subject to at least some high-level control. But the leakage of authority in the whole structure is so great that the pressures felt by low-level officials are often almost totally unrelated to the objectives of those at the top of the pyramid.
In certain respects, bureaucratic free enterprise and extreme ossification have the same impact upon a bureaucracy's overall capabilities. In both cases, top-level officials are unable to accomplish certain high-priority tasks through the normal bureaucratic structure. Either they have almost no control over low-level bureaus (as in bureaucratic free enterprise), or their attempts to impose controls on these bureaus have created complete incapacity to act with dispatch or innovation (as in extreme ossification). As a result, whenever the leaders want to accomplish urgent tasks that require complex operations and considerable innovation, they must set up special organizations outside the normal bureaucracy. These organizations then become subject to all the later phases of the rigidity cycle described above.
Attempts to escape the normal rigidities of a large bureaucracy can take other forms as well. One example is the "campaigning" approach described by Alec Nove in The Soviet Economy.^[Alec Nave, The Soviet Economy (New York: Frederick A. Prager, 1961), p. 289.] Soviet leaders often try to accomplish key economic objectives by launching special "campaigns" that receive intensive publicity for a short period. Just as the privileged status of special organizations tends to dissipate, so the energy invested in each "campaign" gradually declines to a mere trickle. Then a new "campaign" aimed at some other objective is started. This method of attacking key problems outside the normal bureaucratic structure (or stimulating it into abnormal activity) seems to be a response to either ossification or bureaucratic free enterprise, or both.
These phenomena have some implications that are potentially useful in the analysis of communist behavior. For example, if a task is assigned to a new organization endowed with special powers, the task must be considered very important by the leaders. Conversely, if it is assigned to a relatively ossified bureau, it is probably not considered unusually significant. Moreover, special organizations or "campaigns" are almost certain eventually to degenerate to more normal status. Experience might even produce a "normal" time-profile for such degeneration useful in forecasting the future course of specific programs.
Another response to both ossification and bureaucratic free enterprise in large bureaucracies is periodic revamping of control and authority relationships between major components of the bureaucracy. Such reorganizations have four characteristics that create a "cycle":
-
They involve a reshuffling of relationships at higher and intermediate but not at the lowest levels.
-
They are effective in the short run.
-
Any gains in performance are gradually obliterated.
-
Reappearance of the original lack of control eventually stimulates another reorganization.
From the viewpoint of a bureaucracy's top-level officials, periodic reorganizations are designed to influence the behavior of the low-level officials who actually carry out operations. However, top-level officials are rarely able to revamp these lowest levels directly. They usually cannot inform themselves sufficiently about those levels to make sensible suggestions for reform. Also, drastic changes adopted on all levels simultaneously might be too disruptive. This was certainly the experience of the Chinese Communists when they attempted "The Great Leap Forward." Top-level officials usually reshuffle the relationships among major bureau components, but not within each component. Hence such periodic reorganizations rarely penetrate down to the low levels where final action takes place.
Nevertheless, a drastic "shake-up" of intermediate and upper levels may significantly improve the productivity and flexibility of the bureaucracy as a whole. If previously rigid channels of authority and communications are broken up, the bureaucracy will find itself in a relatively fluid state. Such fluidity may release energies and ideas that have previously been held back by ossification. It may also force free enterprising bureaucrats to follow their superiors' orders more closely, until they can determine exactly how much autonomy they now enjoy. Therefore, major reorganizations are typically followed by a short period of chaotic readjustment and a longer period of improved performance.
But this "honeymoon" does not last. The forces that originally caused ossification or free enterprise gradually reassert themselves. Officials set up new channels and networks of authority and communications, both formal and informal. Extensive new rules are developed; and as time passes, these channels, networks, and rules grow more rigid. Moreover, top officials themselves further this process in two ways. First, they soon shift their attention from the reorganization to other matters. Second, they develop new forms of control appropriate to the newly reorganized structure. Eventually the new forms of organization become just as unmanageable as the old ones were. As a result, the leaders' frustrations will gradually rise until they are stimulated to undertake another reorganization. Hence large bureaucracies formed from many component bureaus will be subjected to periodic reorganizations by their sovereigns or topmost officials. Such reorganizations will occur most often in bureaucracies that exhibit the characteristics that make ossification and bureaucratic free enterprise most likely--very large size and weak feedbacks from clientele. One example of such a cycle is Soviet postwar shifts between centralized and decentralized economic controls. Another has been suggested by Robert Perry of The RAND Corporation, who contends that there is a definite seven-year reorganization cycle in US. Air Force procedures for controlling research and development contracts.
Most reorganization cycles do not exhibit consistent periodicity. Nevertheless, the cycle concept could still prove useful in forecasting the behavior of large bureaucratic systems. If definite stages in the cycle of a given system can be identified, then an analyst might indicate the current position of the system in the cycle, predict what the next stage will be, and estimate approximately when it will occur. Admittedly, the operation of such cycles is also significantly influenced by both exogenous events and related long-range trends.