University of Aarhus

1. The Problem

Theoretical questions tend to arise for practical reasons. This is probably why interest in the role of bureaucracy¹ in foreign policy-making has been so slow in developing, for until quite recently there seemed to be no reason for viewing bureaucratic influence in foreign affairs as a practical problem.

As long as foreign policy was considered the prerogative of the executive branch, bureaucracy, such as it was, was expected to share with the political leader or leaders² - who not infrequently were recruited from its ranks - the task of guarding and executing a depoliticized national interest. The principal problem, according to this interpretation, was rather seen to be the incursion of irresponsible party politics into this area. And even when this normative ideal yielded to that of democratic political control of foreign policy, the potential conflict between politicians and bureaucracy could be viewed as the seamy side of the necessary collaboration between these two forces. Perfect political responsiveness and control might be required to retreat somewhat in the interests of bureaucratic efficiency, but this could still be seen as an expression of complementarity, not of bureaucratic encroachment.

As a normative ideal, this view of bureaucracy keeping within its appointed limits is with us still; and even in empirical theory, its various incarnations, most prominent among them the Rational Actor Model,³ have shown great resilience, although developments in the field of organization theory might well have suggested an alternative approach, or at any rate a significant modification of the Rational Actor Model. One reason for this may be the modest size of the foreign affairs departments of most countries, at least until the Second World War; another, the attractiveness in the field of foreign policy analysis, where detailed information about the inner workings of the objects studied are often particularly difficult to obtain, of the Rational Actor Model which, by its intrinsic logic, fills out the gaps in the analyst's knowledge.⁴

---

* An earlier version of this article was presented at the ECPR workshop on 'Sources of Foreign Policy', London, 7-13 April, 1975.

* An earlier version of this article was presented at the ECPR workshop on 'Sources of Foreign Policy', London, 7-13 April, 1975.

The attack on the Rational Actor Model and the awakening of interest in the workings of foreign policy departments seem to a large extent to have been motivated by the tardy realization that bureaucracy might, even in this field, be usurping the functions which ideally ought to belong to the political leadership. Not unnaturally, the objects on which this new discovery was demonstrated or tested were primarily those where the bureaucratic influence was particularly at variance with the presuppositions of normative theory, or the consequences particularly striking.⁵

Against this background, it would seem unreasonably academic not to include the problem of the influence of the bureaucratic 'apparatus' relative to that of the political leader among the central questions which a systematic investigation of the department in foreign policy-making should try to answer. However, it seems fruitful to pose the question in general rather than in particular terms, and neutrally rather than passionately: to ask, not 'Why on earth was the outcome so far removed from the intentions of the political leader in this case, and how far was bureaucracy responsible?', but 'To what extent do the typical workings of foreign policy departments influence the ability of the political leader to perform his role in the manner postulated by the ideal of democratic decision-making and embodied in the Rational Actor Model?' Formulated in this way, the question immediately shows itself to be embedded in the larger problem of how the foreign policy department functions, which must be the second, and indeed main, concern of scholars in this field.

As designated in this abstract manner, the problems may of course be investigated by different approaches, and with quite different kinds of results. The choice of approach should however, in my opinion, be influenced by two major considerations: comparability and realistic operationalization. The idea behind the first of these considerations is evidently that the framework of inquiry should permit and even facilitate cross-national comparison of the results obtained. The second consideration is slightly more complex: it implies that it must be possible to operationalize the variables selected for study in such a way that the questions formulated will stand a fair chance of receiving answers beyond purely formal or trivial ones. The empirical study of politics is, like its subject matter, the art of the possible: and in the case of foreign policy departments, the object of study is likely to prove more recalcitrant than most.

Both these considerations seem to direct the study of departmental functions away from substance, which is often both unique and touchy, and towards the general aspects of process and structure, which will be easier to compare and probably easier to study. This general orientation obviously has its costs, since the answers we receive may be at one remove, at least, from the reality we want to understand; but the benefits seem to be more weighty, particularly since it should not be impossible to bring out points of a substantial nature in terms of more general, structural variables.

Apart from these considerations of comparability and realism, the framework
chosen should of course be such that it predetermines as little as possible the
concrete results to be obtained; nor should it tend to prejudge the normative

issues involved. In both these respects, the option in favour of generality raises
problems which must be briefly discussed.

If we try to answer the question of the functions of departments in general terms, we will tend to highlight the more structural aspects of departmental work, that is, the relatively constant elements of the process observed: the departmental routines, standard operating procedures, etc. In short, questions about the functioning of departments will tend to produce answers about departmental sluggishness and inertia. A bureaucracy portrayed in such terms would prima facie seem to play not the flexible, subservient role vis-å-vis its political leader which normative theory assigns to it and the Rational Actor Model assumes, but to stand instead for a constant narrowing of horizon and frustration of purpose. Thus, the question of the influence of the department on the political leader would apparently be predetermined by the choice of a general and structural description of the bureaucracy in question.

This methodological difficulty probably cannot be entirely avoided, but it should be stressed that departmental structure may well to a significant extent turn out to consist of procedures which, though 'standard' and unconscious, are conducive to some kind of rational action,⁶ even though the rationality will hardly be full. The identification of similar structures in the behaviour of the political leader himself would tend to lend added weight to this observation.

Certain normative consequences also flow from the adoption of a general approach. Describing departments in terms of structures means that the most prominent part of the description will cover those parts of behaviour which are most permanent and general. But the greater the permanency of the features, the more difficult will it become to draw normative implications. If the characteristics of the 'machine' which are the causes of its (normatively) undesirable performance can be shown to be ineluctable, all genuine responsibility will of course disappear ⁷ ultra posse nemo obligatur. The possibility mentioned above that the 'pathological' characteristics might be universally present, not only within bureaucracies but among the political leaders themselves, would of course logically only reinforce this line of thinking. Empirical science can show no way out of this dilemma, since it cannot tailor its substantive results to fit normative concerns. But those who fear the advent of absolute determinism in this field may take heart from the general experience according to which responsibility will often be made to stick in practice in cases where the theorist would see only disculpating necessity.⁸

2. The Field

Before we move on to a more detailed discussion of what framework to adopt for
analysing the functions and influence of bureaucracy in foreign policy, it may be
useful to survey the work already done in this field.

A basic distinction can be made between those studies which rely on an explicit
conceptual framework, or are at least written from a theoretical perspective, and

those which lack this theoretical dimension. Among the latter, much of the material is in the form of essays and memoirs by diplomats and politicians, works which, while often rich in insights and reflections, remain impressionistic and unsupported by hard data.⁹ Another group is that of studies elaborated as a basis for improving the efficiency - variously defined - of foreign policy bureaucracies.¹⁰ In some of these studies, the normative goal predominates over more analytical considerations, although much potentially useful information may often be found;¹¹ other works in this group, however, attain a level of methodological sophistication which translates them into the loftier - and more useful - sphere of policy science.¹²

It is of course within the category of theoretically 'conscious' works that we should look for real help in reflecting on a useful approach to the study of bureaucracy in foreign affairs. Very naturally, the framework adopted by the studies in this category is borrowed from theories of organizations, large organizations in particular. However, in the adaptation of these theories to their specific purpose, they show considerable variation, and present similarly varied difficulties.

Some works are written within the traditional framework of foreign policy decision-making theory first outlined by Snyder.¹³ This certainly brings the advantage of a systematic approach to the subject; and the central premiss of decision-making theory according to which 'foreign policy-making is most fruitfully analyzed as decision-making in an organizational context'¹⁴ would seem to make this approach an obvious point of departure. However, the decisionmaking approach is, from our point of view, subject to a number of limitations. First, it is obviously elaborated with a view to the analysis of the processes leading to major foreign policy decisions - the more important and critical the decision, the better. Conversely, the framework is less suited to the study of the iterative phenomena which make up organizational structure. Secondly, the approach does not specifically focus on the analysis of organizations, but is designed to cover all the elements in the decision-making process. And finally, the concept of organizational unit is defined within the approach so that it often refers to the small, select circle of decision-makers upon which interest tends to focus - the real leaders in fact - while our object of study, the department, is in fact simply regarded as 'a great pool of personnel and information for the decisional unit.'¹⁵

The bureaucratic politics model, as originally elaborated by Allison and developed in a number of later statements,¹⁶ has been gaining ground rapidly in recent years. There is no doubt that its wealth of variables makes it more 'realistic' than other and more simplified models; and its main prop, the idea of intraorganizational 'pulling and hauling', has an immediate appeal to most persons with bureaucratic experience. Moreover, this central conception would seem to indicate that the bureaucratic politics approach was directly relevant to the question of the possible usurpation by bureaucracy of the functions of political leadership.

Here again, however, one comes to recognize that the approach suffers from
a number of limitations. The bureaucratic politics model, to an even greater de

gree than the decision-making approach, is designed to aid the analysis of the processes leading to specific political outcomes: what it asks questions about are the concrete players, stands, etc. (i.e. factors which vary from case to case). Admittedly, the bureaucratic politics theorists themselves try to formulate general propositions, but these often take the form of precepts for single, rational players in the bureaucratic politics game, and cannot readily be aggregated to yield hypotheses about the results of the overall game under various conditions.¹⁷ The central problem in trying to generalize from the bureaucratic politics paradigm is probably that expressed in Allison's dictum that 'the hard core of the bureaucratic politics mix is personality'.¹⁸ From the point of view of theorists who wish to be able to cumulate knowledge, this hard core melts into a soft centre.

Apart from these problems, the bureaucratic politics approach labours under the special difficulty that the evidence which it requires is, in most political systems, exceedingly hard to get: it cannot be realistically operationalized. Not only does this mean that the intuitive appeal of the paradigm cannot in practice be substantiated by reliable data except in political systems as open as that of the USA, but it is by no means certain that evidence from more closed systems, as for instance those of Continental Europe, had it been available, would in fact have supported the bureaucratic politics hypotheses to anything like the degree which we find in some of the American cases discussed by Allison, Halperin and others. The administrative tradition behind the secretiveness which frustrates bureaucratic politics scholars in search of material may well also be of a nature which inhibits pulling-and-hauling at the bureaucratic level.

The cybernetic-cognitive model of decision-making elaborated by Steinbruner¹⁹ is interesting for its general attempt to link up organization theory with cybernetics ²⁰ But Steinbruner's distinction between the traditional, 'analytic' paradigm characterized by conscious, if limited, value integration, and the paradigm of unconscious, cybernetic adaptation is not convincing when applied to organizations whose genesis and growth are by no means simply or even predominantly 'organic', but are to a large degree the result of more or less successful planning. Because of this comparative fragility, and a certain general vagueness, of the central assumptions of Steinbruner's paradigms, they remain an inspiration rather than a model to emulate.²¹

The most fruitful existing theoretical framework for our purposes would seem to be the organization theory of Simon, March and Cyert, and others.²² Among the most important studies of departments in foreign policy which place themselves squarely on this basis are William Bacchus's Foreign Policy and the Bureaucratic Process²³ and parts of Allison's Essence of Decision (the Organizational Process Model). Here, the centre of interest is definitely the bureaucratic organization; the theory arms at delineating types of administrative structures and process, not at providing concrete explanations and predictions, and the main assumptions are precise and fairly well consolidated.

Even this approach, however, is not without its difficulties when confronted with
our concerns. One such difficulty is the curious resistance exhibited by classical
organization theory when attempts are made to extract useful and testable hypotheses

from it: a certain amount of redefinition is often needed. The most consistent attempt to overcome this difficulty and apply organization theory to questions of bureaucratic functions and influence in foreign policy-making is probably Allison's above-mentioned work. But even this cannot in my opinion be taken as a guide without a certain amount of qualifications, which all stern from the fact that Allison's models are primarily intended to provide a basis for the explanation of specific occurrences in the field of foreign policy.

A first consequence of this, and of the fact that the examples Allison discusses are taken from a game of high politics played between the superpowers and with strong military components, is that Allison, in stressing the role of routines, standard operating procedures and the like, is to a considerable extent concerned not with the production of decisions, but with their physical implementation. However, a large number of the products of foreign policy-making, especially in countries with more limited roles and ambitions in the international system, are either self-implementing, like some official statements of position, or have to be implemented in verbal rather then physical terms. This means that the problem of the influence of bureaucracy on output takes on a complexion which is rather different from that of the questions treated by Allison.

Secondly, the elements of the paradigm which Allison elaborates in connexion with his Organizational Process Model are oriented more towards substance than towards process and structure. An inquiry along the lines he suggests is more likely to produce interesting answers about what premisses the bureaucracy came up with in a particular case, than how it came up with them.

And finally, the orientation towards specific explanation pushes Allison to formulate a 'dominant inference pattern' for his Organizational Process Model according to which '[the] Model's explanatory power is achieved by uncovering the organizational routines and repertoires that produced the outputs that comprise the puzzling occurrence.'²⁴ This may be didactically useful as the embodiment of a significant aspect of explanation, but it definitely seems to me to overstress the role of bureaucratic routines.²⁵ If we are to remain faithful to our option in favour of generality, it seems more fruitful to view bureaucratic influence not as the sole or main determinant of individual cases but as a general constraint, of varying magnitude, upon classes of output: Our main question does not so much deal with whether certain bureaucratic factors can actually be said to have caused the puzzling occurrence X, as with the structural bureaucratic constraints which can be discerned and which will typically tend to affect foreign policy-making or certain kinds of foreign policy-making.

3. The Framework

After this rapid review of the field, I shall now try, first, briefly to define the outlines of a theoretical framework for the study of bureaucracy in foreign policymaking which may have certain advantages with respect to generality and operationalization, and then to show in more concrete terms what questions may

be formulated and what hypotheses tested along the lines suggested by this framework.

It seems useful to define the dependent variable of the study in the manner of organization theory, i.e. as 'output' instead of as 'decision'.²⁶ The notion of output, as opposed to that of decision, is neutral with regard to the character of the process leading towards it, and thus implies no bias in favour of rationality or importance.²⁷ This choice of dependent variable therefore covers the large class of activities such as negotiations, international conferences, speeches and statements, travels, etc., which play a major part in the life of foreign policy departments, but which are not easily accommodated under the heading of 'decision'.²⁸

In more concrete terms, the 'output' of an organization will here be defined as a communication from the organization to the environment which in some sense formally involves the organization as the source of information or as taking a stand on some issue. The central notion is that of 'engaging the responsibility' of the organization and, in the case of government departments, of the country whose interests it serves and represents.²⁹

In itself, this definition only seems to shift the question. Output is viewed as communication, but communication of what? If the answer turns out to be 'of decisions', there seems to be no need for the theoretical convolutions of organization theory.

A way out of this difficulty is offered by the idea proposed by Niklas Luhman ³⁰ that 'decision' be defined as 'communication of the result of a processing of information'.³¹ This processing is twofold: on the one hand, the decisionmaker reduces complexity, that is, he passes on less information than he has received. This is the essence of the internal processing of a particular question. And on the other hand, the 'decison-maker'³² 'passes on more information than he has received, and in this sense assumes responsibility.'³³ This surplus of information is in fact identical with what one might otherwise call the decisional element, since the actual production of a particular output or 'decision' by an organization can never be deduced from the premisses, if these are viewed simply as informational input.

In this model, the 'value premisses' and 'factual premisses' of conventional decision and organization theory are thus both assimilated into the general concept of information. If the chain of reasoning behind this assimilation is accepted, information seems to be a promising central organizing variable in the study of departmental process. The various inputs and withinputs can all be viewed in their informational aspect, and the process itself as a continual reception of, search for, processing and transmission of information.³⁴

The study of organizations in terms of information can move on and between different levels of abstraction; even on the highest level, which deals with sources and channels of information, and general factors of 'noise' - objects of study which should prove acceptable even to fairly closed bureaucracies - the investigation can probably yield results of substantive interest. To this general advantage should be added the specific one that information seems particularly appropriate as an organizing variable in the field of foreign policy, where reliable

knowledge of the environment is a particularly scarce resource, both for departments
and for the general public, and where so much output takes the form of
purely verbal behaviour.35

The concept of information is not, however, sufficient as an organizing variable. The actual character of the gathering, processing, and transmission of information in the organization will in many cases be determined by the need for the result of the process to be transmitted to the environment as organizational output. This means that the process will be affected by the presence or absence of deadlines for the production of such output or for the various stages of the internal processing, that is, by considerations of time.³⁶ This concept should therefore be added to that of information as an organizing variable.

It may be noted that the two organizing variables which we have selected are closely connected with the variables in terms of which attempts have been made to construct a typology of situations in foreign policy decision-making. Charles Hermann, in his studies of crises in foreign policy,³⁷ distinguishes three dimensions: threat, anticipation, and time for decision, a choice of situational variables which seems to command fairly wide acceptance. Since anticipation of an event is heavily dependent on previous knowledge, the parallel with respect to both the information and the time variable may be said to be perfect.

A case can be made for viewing the 'threat' dimension as the essential crisis variable.³⁸ But it should be stressed that this still means that the information and time variables can take on a number of different values both in crisis₃₉ and in non-crisis situations, and does not in any way imply that the study of these aspects of organization will deal with particularly 'routine' features, in Hermann's sense of the word.

4. The Questions

Above, the general argument was put forward that the concepts of information and time are a suggestive basis for the study of the functions and influence of bureaucracy in foreign policy-making. Below, I shall try to show in concrete terms that it is possible on this basis to delineate promising areas of study, and to formulate fruitful questions and hypotheses.

In this latter respect, I have not felt that any of the existing approaches, including the classical organization theory, which is the central source of inspiration of the framework outlined above, by itself offered a sufficient number of interesting and testable propositions.⁴⁰ Nor have I, at this point, felt it possible to elaborate a formalized model to be tested against the empirical material. The source of the hypotheses will therefore be quite varied, and cover not only theoretical works in various fields, but also, to a significant extent, impressions and intuitions of participant observers, including my personal experience in the Danish Foreign Ministry.

This particular national background may of course have had the effect that
certain of such propositions will not be borne out in different national systems
or different types of system. In fact, though, I suspect the opposite to be true:

that elements which I have weeded out from the hypotheses for being, prima
facie, too exclusively Danish in their character may turn out, if tested, to be
quite generally applicable.

While the focus of attention has hitherto been on the organizational output, it
seems natural to begin the survey of concrete questions and hypotheses at the
other end of the process, i.e., on the input side.

Departmental function

Members of an organization usually possess a fund of information⁴¹ which, though relevant to their organizational activity, is tied to the person rather than to the organizational function which he fulfils. This personal 'informational background' of organization members will include both pre-entry training and experience and post-entry training, career experience, and acquisition of general information of relevance to organizational activity. It is worth studying this information in some detail, although its actual influence on organizational process cannot of course be deduced from its mere existence, which only indicates potential influence.⁴²

Turning from personal to organizational information, all departments may be presumed to receive informational input which is unrelated to the needs of any specific problem defined in advance. This is probably true of some of the reporting of diplomatic missions abroad,⁴³ the routine scanning of newspapers and periodicals, and such meetings with persons from other departments, politicians, persons from local diplomatic missions, and private members of the public which are designed and expected to yield only generally relevant information.⁴⁴

The questions in this area may include the amount of general information received, its sources, its nature and the circle of recipients. The question of sources and recipients is a particularly interesting one because a chart of information flows may provide empirical indications of some of the cleavages and conflicts within the organization described by the bureaucratic politics approach: The withholding and volunteering of information is a powerful weapon in the hands of organizational subunits, since influence on decisions depends on knowing when and where they are discussed and in what terms.⁴⁵

The hypotheses in this area include the following: General information will account for an appreciable part of the total information received. General reporting of missions abroad will often be based on press reports or other previously published material;⁴⁶ the contents of such reporting will mostly be relevant to current problems and not to long-term considerations. The circle of recipients within the department will be wide. The extent of direct contact between officials and politicians will depend on the political system in question.⁴⁷

The category of ad hoc information, as distinct from that of general information, presupposes the existence of a concretely defined need for output which sparks off information search and processing. Such search and processing guided by a concrete output focus will be called a 'case'.⁴⁸

The initiation of a case includes a time element in the form of the setting or

anticipation of some deadline. Formally, the initiative will always come from within the ministry, since public bureaucracies are not supposed to be 'on tap' for direct outside initiatives. On the other hand, this formally internal initiative will almost invariably be connected with some outside pressure. This pressure may be a direct inquiry, complaint or request from the environment, or it may be the communication of dates for future conferences, visits, negotiations, etc., particularly if accompanied by an indication of the agenda. The pressure resulting in the initiation of a case may exist before the case is started, or it may have been anticipated by the minister or members of the department. In either event, the question of initiative is important, because it not only defines time limits⁴⁹ but also contributes to the definition of the problem to be solved by placing it in a particular structural setting. For instance, it may determine through what 'action channel' the case is going to proceed.⁵⁰

The main questions to ask in this connexion seem to concern the kind of events that stimulate the initiation of a case, and the more or less active nature of the organization, i.e., how much environmental 'prodding' is required to start a process towards an output.

A reasonable hypothesis in this respect would be that the department will anticipate structural events (e.g., scheduled meetings) much more readily and consistently than hypothetical, even though very likely, chains of events leading to possible problems.⁵¹ In general terms, it is to be supposed that organizations have a predilection for the actual over and against the hypothetical.

Once the case is initiated, it will follow a more or less predetermined route within the organization. Relevant questions in this respect may include where this route starts, how many separate hierarchical levels it will typically pass through, what levels may be bypassed,⁵² and at what level the final authorization of output is given.

The next point, which is of special interest, is that of the information used in a particular case. First, what is the balance between the general information already present in the organization and the ad hoc information collected or volunteered because of its direct bearing on the case in hand? And secondly, what kinds of general and ad hoc information are most commonly used in the various cases? Concerning ad hoc information, further questions might try to elicit what are the typical sources of such information, whether it is proffered or has to be solicited by the organization, and what procedures of search are typically set in motion.

Reasonable hypotheses in this area might include the following: General information is used to a much lower extent than that suggested by its total bulk. Ad hoc information, however, will also be sought only to a fairly limited extent.⁵³ Procedures of search will typically be fairly stereotyped and tend to remain in the immediate vicinity of already existing solutions to analogous problems.⁵⁴ An important item of information in multilateral settings concerns the stands of other nations and international organizations involved.⁵⁵ It also seems likely that ad hoc information will usually be sought in integrated form, i.e., without overt attempts at separation into factual and valuational components.

As we have seen, information is not totally given for an organization but may be influenced by procedures and decisions. This is true, however, not only of the information but also of the time dimension. As we have seen, the precise time at which a case is initiated to a large degree depends on decisions within the ministry. Even after the constitution of a new case, however, there is latitude for further organizational manipulation of the time factor. The existence of a new case does not always in itself imply that a definite deadline, a concrete terminus ad quern, has been laid down. And even where this has occurred, the terminus a quo, i.e. the moment at which actual search processes, etc., are started, is not predetermined. It therefore seems useful to try to pin down the typical actual time limits of work on given cases.

The hypothesis which seems most likely to be substantiated in this respect is perhaps that actual processing time is very much shorter than potential processing time, or, to be more concrete, that work on given cases will typically begin from 24 to 72 hours before the imposed deadline.

A number of arguments may be adduced in support of this hypothesis. For instance, experiments in small group theory⁵⁶ seem to show that a mild degree of stress is eufunctional with respect to problem solving.⁵⁷ An observed time span in the neighbourhood of the one postulated may, moreover, be related to the hypothesis concerning the balance of general and ad hoc information, and that concerning the volume of the latter category, the influence of the second proposition reinforcing that of the first one towards a delay of processing until shortly before the output deadline. This accords well with the generally noted predisposition of organizations in favour of shortrun feedback in order to reduce uncertainty.⁵⁸

Departmental influence

It seems reasonable to suppose that an inquiry along the lines set out above would go some way towards providing an answer to the question concerning the functions of departments in foreign policy-making. When we turn to the question of the influence of the department relative to that of the minister, one factor of explanation must, within the present framework of inquiry, be discounted at the outset, namely that of the personality of the minister and of the departmental chiefs. As I have tried to argue, factors of this kind, although they may be extremely important in practice, are almost impossible to treat in general terms without losing the major part of their interest.

Apart from this, the influence of the minister on output seems in the main to be tied to three factors: he must be able to communicate his ideas to those levels of the department where the output is prepared and authorized; he must have sufficient acquaintance with the actual cases treated in the organization to know where he wants to make his influence felt; and he must be able to control in practice the nature of the output on which he has wished to exercise this influence.

The first factor raises the so-called 'guidance problem'. The questions which

may fruitfully be asked in this connexion seem to include the following: In what form does the minister communicate his 'guidance', i.e. his views? If it is written, to whom is it circulated, and how quickly? If it is oral, how is its further transmission assured? How often does the minister meet with officials of the department? With whom does he meet? What other sources of direct contact with officials does he have?⁵⁹

The underlying hypotheses in this area are mainly those elaborated by information
theory concerning factors of distortion of information.60

The second factor mentioned above concerns the possibility for the minister of acquainting himself with the potential areas of his influence, that is, the cases being treated in the department. He may come to know of these cases either on his own initiative or by having them submitted by officials of the department. In the former case, the question is to what extent the time schedule and the structure of contacts of the minister permit him to follow the work of the department off his own bat. A likely hypothesis would here be that such general surveillance would normally be slight, sporadic, and unsystematic.⁶¹

When officials submit cases, the question is of course what types of cases and how many will be submitted. The hypothesis could in this respect be based on the findings according to which crisis situations draw the locus of decision upwards in the hierarchy. If we accept the argument referred to earlier that the 'essential' crisis variable is that of threat (i.e. of high priority interests), the cases presented for ministerial decision would presumably be those that involved central interests ⁶² while even urgent cases would tend to be dealt with at lower levels if they only affected lower priority interests. This would be true even if previous guidance on similar cases was slight. The importance of empirical findings in this area is obvious.

The third factor is that of the possibility of effective influence and control.
Here again, questions concerning time and information may highlight central
features of the problem.

The time dimension is of course again partly dependent on the time schedule of the minister.⁶³ But it is also affected by the time given him by the department to consider the cases submitted, and by the form in which the information defining the case is submitted.⁶⁴ In other words, how long before externally imposed deadlines will the minister typically be able to review a case, and to what extent does the presentation of the case leave him with an effective choice, within the constraints of time, between alternatives?

The hypothesis would here be that the effective time for review would be fairly short (partly as a result of the general tendency towards 'eufunctional urgency' noted above), and that the case would normally be presented in a unitary' form, i.e. as a single recommendation supported by arguments.⁶⁵ A possible countervailing tendency would be the oral discussion of cases between the minister and a number of officials of the department, often in situations where time was particularly short; such a procedure would probably allow for more effective discussion of alternative views,⁶⁶ but the time factor militating in favour of oral discussion would at the same time tend to cut wide-ranging discussion short.

5. The Prospects

In order to carry out an investigation of the kind sketched out above, a variety of means could be used. While a prominent place should be given to questionnaires and quantitative measurement of such variables as show themselves amenable to that treatment, the collection of 'semi-soft' data in the form of structured reporting by selected participant observers or observant participants would be extremely useful,⁶⁷ as would also the 'soft data' provided by freewheeling, openended interviews, especially with high-level participants.

It would probably be prudent to postpone international comparison until a national survey in depth had been carried out and evaluated. After that, of course, comparison not only in space but also in time, i.e. by means of longitudinal research into the same organization, would become an obvious necessity.

---

NOTES

1. For all its unfortunate normative overtones, the concept of bureaucracy is too useful to be dispensed with in the discussion. Later on, the object of study will, for convenience, often be referred to as the 'department', 'departmental process', etc. The foreign policy relevance is assumed throughout, even in the absence of specific formulations indicating it.

2. This concept covers not only ministers, but also other top-level political appointees. For the word 'minister' will be used to denote the whole

3. Cf. Graham T. Allison, Essence of Decision (Boston: Little, Brown, 1971), pp. 10-38. In one sense, of course, the Rational Actor Model is silent on the role of bureaucracy; but this silence logically implies that bureaucracy does not play a role in the elaboration of policy which makes it necessary to adjust the goal presuppositions of the model in order to predict output correctly: in a monolith, all the elements are supposed to be in perfect harmony.

4. A third factor, which should not be ignored, is the deliberate wish by expositors to portray the foreign policy process in a way as close as possible to the democratic ideal. While the reasons for such a course are readily discernible and quite understandable in the case of actual participants in the process [see, for a good example, the article on the Danish Foreign Service by Torben Rønne in Fremtiden, vol. 26 (1971), pp. 12-15), it is a little disturbing to find similar inclinations within the academic community (cf. Stephen Krasner, 'Are Bureaucracies Important? (Or Allison Wonderland)', Foreign Policy, 1972, pp. 159-79, whose double-barrelled - normative and empirical - criticism of Allison, op. cit., may well be read to imply that scholars ought to continue to portray the making of foreign policy as being compatible with democratic ideals, regardless of whether empirical research is able to back up such a description].

5. Prominent examples are the Bay of Pigs Affair 1961, the Cuban Missile Crisis 1962, and the Vietnam War. Cf. Allison, op. cit., and Irving L. Janis, Victims of Groupthink (Boston: Houghton Miff lin, 1972).

6. This is particularly evident if one introduces economy among the goals and information search among the cost factors of such rational action, cf. Julian Feldman & Herschel E. Kanter, 'Organizational Decision-Making', James G. March (ed.), Handbook of Organizations (Chicago: Rand McNally, 1965), pp. 637-38.

7. Cf. Krasner, op. cit., pp. 160-61,

8. See, for one example, the fate of Admiral Kimmel after Pearl Harbor. Roberta Wohlstetter may well argue that 'incredible as it may seem, Admiral Kimmel's defense was much closer to the truth than the accusations leveled against him' (quoted in Janis, op. cit., p. 84); he was demoted all the same. For the view that bureaucracy has in fact unjustifiably become a scapegoat, cf. Francis E. Rourke, Bureaucracy and Foreign Policy (Baltimore/London: Johns Hopkins University Press, 1972), pp. 6-7.

---

8. See, for one example, the fate of Admiral Kimmel after Pearl Harbor. Roberta Wohlstetter may well argue that 'incredible as it may seem, Admiral Kimmel's defense was much closer to the truth than the accusations leveled against him' (quoted in Janis, op. cit., p. 84); he was demoted all the same. For the view that bureaucracy has in fact unjustifiably become a scapegoat, cf. Francis E. Rourke, Bureaucracy and Foreign Policy (Baltimore/London: Johns Hopkins University Press, 1972), pp. 6-7.

9. Among the more useful can be mentioned, at random, Lord Strang, Home and Abroad (London: Deutsch, 1956) and Ivone Kirkpatrick, The Inner Circle (London/New York: Macmillan, 1959).

10. For a survey of efforts in the United States in this direction, cf. Irving M. Destier, Presidents, Bureaucrats and Foreign Policy. The Politics of Organizational Reform (Princeton: Princeton University Press, 1972).

11. See, for instance, Diplomacy for the 70's. A Program of Management Reform for the Department of State (Washington: U.S. Government Printing Office, 1970).

12. See, for instance, Burton M. Sapin, The Making of United States Foreign Policy (Washington: Brookings Institution, 1966) and Alexander L. George, 'The Case for Multiple Advocacy in Making Foreign Policy', American Political Science Review 66 (1972), pp. 751-85.

13. Richard C. Snyder, H. W. Bruck & Burton M. Sapin (eds.), Foreign Policy Decision (New York: Free Press, 1962). For a later exposition, see James A. Robinson & Richard C. Snyder, 'Decision-Making in International Polities', in Herbert C. Kelman (ed.), International Behavior: A Social-Psychological Analysis (New York: Holt, Rinehart & Winston, 1965).

14. et al., 1962, op. cit., p. 103.

Snyder 15. Ibid., p. 98.

16. Allison, op. cit., pp. 144-84; Graham T. Allison & Morton H. Halperin, 'Bureaucratic Politics: A Paradigm and Some Policy Implications', World Politics, 24 (Supplement 1972), pp. 40-79; Morton H. Halperin, Priscilla Clapp & Arnold Kanter, Bureaucratic Politics and Foreign Policy (Washington: Brookings Institution, 1974).

17. See, for instance, the 11 + 9 general rules for affecting information in a manner favourable to oneself in Halperin et al., op. cit., ch. 9. Such precepts are reminiscent of the propositions in Anthony Downs, Inside Bureaucracy (Boston: Little, Brown, 1967), but lack the stringent fundamental assumptions of the latter work, and in fact sometimes do not seem to add materially to the results of unaided reflection on the reader's part. Possibly, the bureaucratic politics paradigm would not be very much better off if its individualist propositions could in fact be aggregated: cf. the critique of Robert J. Art, 'Bureaucratic Politics and American Foreign Policy', Policy Sciences, 4 (1973), pp. 467-90, where the point is made that the final, unintended resultant from the pulling and hauling may not differ materially from that which the intentions of the participants would indicate.

18. Allison, op. cit., p. 166.

19. John D. Steinbruner, The Cybernetic Theory of Decision (Princeton: Princeton University Press, 1974).

20. A particularly useful feature of this translation seems to me to be the way in which the organizational character of decision-making is placed not as a presupposition, simply defining decision-making problems as problems of organization theory, but as one of the elements of complexity surrounding the decision-makers.

21. Of especial interest is Steinbruner's attempt to formulate the seemingly very individualoriented questions suggested by the inclusion of cognitive psychology, within a context in which organizational factors are postulated as intervening variables. For a somewhat similar, and similarly suggestive, formulation, see Raymond A. Bauer, Ithiel de Sola Pool & Lewis A. Dexter, American Business and Public Policy (New York: Atherton Press, 1963), pp. 142-43.

22. See, for instance, Herbert A. Simon, Administrative Behavior (2nd ed.) (New York: The Free Press, 1957); J. G. March & Herbert A. Simon, Organizations (New York: John Wiley, 1958); R. M. Cyert & J. G. March, A Behavioral Theory of the Firm (Englewood Cliffs: Prentice-Hall, 1963). The theory of 'incrementalist' decision-making of Charles Lindblom belongs within the same tradition.

23. (Princeton: Princeton University Press, 1974). This work, while admirably detailed and systematic in its treatment of a typically 'departmental' area, does seem to suffer from a certain lack of connexion between the theoretical introduction and the main, empirical part.

24. Allison, op. cit., p. 88.

---

25. Significantly, Allison's Organizational Process Model has since merged with the personality-oriented' Governmental Politics Model to form the bureaucratic politics approach in its most recent statement. As I have tried to show above, the role of the latter approach is precisely to produce realistic explanations of specific political outcomes; the bureaucratic politics paradigm may in fact be seen as a gigantic plan for multiple concrete inference patterns - but not for a single general one.

26. I have retained the notion of output in the terminology of the rest of this article. The occasional use of 'decision' and its compounds for stylistic reasons should not be taken to imply a departure from the definition adopted here unless the context plainly indicates the contrary.

27. This neutralization also implies a blurring of the distinction between valuational and factual premisses which is reflected in the idea, discussed below, of accommodating both these categories within the concept of information.

28. Cf. George Modelski, A Theory of Foreign Policy (London/Dunmow: Pall Mall Press, 1962), p. 14.

29. Cf. ibid., p. 5.

30. Theorie der Verwaltungswissenschaft (Cologne/Berlin: Grote, 1966), p. 69.

31. Ibid.

32. In Luhmann's terminology, this concept may be read as identical with the organization from which output emerges, since the voluntaristic aspect of decision-making is conspicuously absent from his model. Cf. op. cit., pp. 69-70.

33. Luhmann, op. cit., p. 69 (my italics).

34. The central role of information is almost universally acknowledged in literature on foreign policy-making. Nevertheless, I have found no major study which specifically centres on this variable for the special purpose of investigating bureaucracy. It is true that Bauer, Pool & Dexter (op. cit.) concentrate on informational factors, but they concern themselves mainly with certain pressure groups and with Congress. The cybernetic theories of Karl Deutsch [The Nerves of Government (2nd ed., New York: The Free Press, 1966)], although clearly relevant to these considerations, seem to me too general in their scope and propositions to serve as more than an inspiration.

35. This is less true of certain kinds of foreign policy activity dealing with economic or commercial interests; it is also less true of systems which are large or independent enough to include physical actions in their standard repertory of diplomatic instruments. In all these cases, however, information remains central for processes leading up to the choice of output.

36. Some deadlines may in fact refer to the production of output which does not reach the environment. The inclusion of such 'intermediate outputs' under the category of regular output would, however, not only entail a shift in the level of analysis, but also create a number of other difficulties.

37. Charles F. Hermann, Crises in Foreign Policy (Indianapolis: Bobbs-Merrill, 1969), p. 36.

38. A 'low threat' situation characterized by short decision time and/or great surprise might well lead to 'ant-hill' behaviour in the organization; but chances would be that the busy hurrying to and fro, telephoning and drafting, etc., thus observed would be described by the participants as simple 'pressure of work', and, if anything, be associated with more 'routine' than other situations in the sense that such pressure might tend to increase the usefulness of standard operating procedures (to save time) and that final decisions might be taken at lower levels (again to save time).

39. For a number of hypotheses concerning communication and information in crisis situations, cf. Charles F. Hermann, 'Some Consequences of Crisis Which Limit the Viability of Organizations', Administrative Science Quarterly, 8 (1963-64), pp. 61-82.

40. Even the study by Ørvik et al. [Nils Ørvik, A. Sjaastad, A. Hallenstvedt, O. Bruaas, J. Jølle, Departmental Decision-Making (Oslo: Universitetsforlaget, 1972)], which explicitly places itself upon the firm ground of organization and decision-making theory, in fact only to a limited extent seems to have profited from this theoretical underpinning when it comes to the suggestion of interesting hypotheses.

41. This concept should still be read as covering both facts and values.

42. Of course, since officials' brains are not neatly compartmentalized into relevant and irrelevant background, this category of information will shade over into the general personality and attitudes of organization members. There is no reason why data concerning such factors should not be regarded as 'information' and included among the aims of empirical research from this perspective, as long as it is realized that this factor may be more difficult to generalize, and its relation to the dependent variable perhaps more problematic, than in the case of organizationally relevant information.

---

42. Of course, since officials' brains are not neatly compartmentalized into relevant and irrelevant background, this category of information will shade over into the general personality and attitudes of organization members. There is no reason why data concerning such factors should not be regarded as 'information' and included among the aims of empirical research from this perspective, as long as it is realized that this factor may be more difficult to generalize, and its relation to the dependent variable perhaps more problematic, than in the case of organizationally relevant information.

43. Such missions may of course be viewed as part of the organization, so that their reporting becomes a 'withinput' rather than an input. In view of the geographical separation and the ensuing difficulties of communication, which may in some instances assume critical proportions, it seems more fruitful to treat the missions as being outside the organization, although of course in a special relationship to it, and themselves a promising area of research along the same lines.

44. Much of this general informational input is of course in fact sought or volunteered in anticipation of specific problems to which it is expected to be relevant. Consequently, the distinction between 'general' and 'ad hoc' information may be seen as one of degree only, and should not be stressed too much. Nevertheless, the ad hoc process of search will tend to exhibit characteristics rather different from general informational processes.

45. Ørvik et al. (op. cit., p. 103) surmise that 'a high sent percentage indicates a high degree of influence over the relevant environment while a high received percentage indicates lack of influence.' This does not seem very convincing: orders may flow from formal centres of influence, but facts will tend to flow in the opposite direction, and it seems impossible to decide whether one or the other category of messages has the strongest influence on output. Cf. Harold Guetzkow, 'Communications in Organizations', in March (ed.), op. cit., p. 539.

46. For a somewhat different view by an observant participant, cf. Arthur Andren, 'His Ex or Telex', International Journal 25 (1969-70), pp. 676-677.

47. Such contacts are usually viewed with a sterner eye in Continental countries than in the United States. Compare for instance Ezra N. Suleiman, Politics, Power, and Bureaucracy in France (Princeton: Princeton University Press, 1974), pp. 298, 311-12 with the information given by G. Russell Pipe, 'Congressional Liaison: The Executive Branch Consolidates its Relations with Congress', Public Administration Review, 26 (1966), pp. 14-24.

48. This term is admittedly not a very happy one, but its possible alternatives are even worse.

49. For a discussion of this aspect, cf. Bacchus, op. cit., pp. 182-83.

50. Cf. Allison, op. cit., pp. 169-70.

51. Bacchus, op. cit., pp. 96-100 reaches similar conclusions.

52. This and the preceding question are relevant to the consideration (Guetzkow, op. cit., p. 537) that time-lags in communication may lead to problems of co-ordination.

53. C/. Hermann, op. cit., pp. 158, 161.

54. This assumption is central in models of organizational process. Empirical support can be found, for instance, in Bacchus, op. cit., p. 185.

55. In a small country like Denmark, this information is important both for its factual content and as a possible 'value' premiss, since it may be useful in indicating that other participants, who are either more powerful or with whom it is otherwise deemed acceptable or even advisable for Denmark to associate herself, are going to take stands which lie near enough to the Danish one for Denmark to abstain from elaborating and advocating a separate policy and simply to shelter behind the nearby one. The 'search economic' value of such information may be very considerable. For a discussion of this point, cf. Johan P. Olsen, 'Public Policy-Making and Theories of Organizational Choice', Scandinavian Political Studies 7/1972, p. 49.

56. See the report and references in Charles F. Hermann (ed.), International Crises (New York: The Free Press, 1972), p. 34.

57. It should be borne in mind, however, that the actual time limits and other circumstances observed in the department may be such as to qualify as 'moderate' or even 'severe' stress, which is seen as dysfunctional.

58. Cf. Feldman & Kanter, op. cit., p. 627.

59. Ørvik et al. (op. cit., pp. 108-21) are particularly thorough in their treatment of this point.

60. Cf. Guetzkow, op. cit., pp. 550-58.

61. Cf. Ørvik et al., op. cit., p. 133. The existence, as in the United States and France, of a staff specially entrusted by the minister with this task may, but need not, alter this pro position. In any case, access to information remains a crucial variable (cf. George, op. cit., pp. 775, 777, for examples).

---

61. Cf. Ørvik et al., op. cit., p. 133. The existence, as in the United States and France, of a staff specially entrusted by the minister with this task may, but need not, alter this pro position. In any case, access to information remains a crucial variable (cf. George, op. cit., pp. 775, 777, for examples).

62. Of course, the minister's perception of what constitutes 'threat' or 'high priority interest' in a given case may not be the same as that prevalent in the department. The extent of departmental learning on this point is an obvious area of study. C/. Rourke, op. cit., p. 65.

63. This point, and its connexion with the provisions governing the relations between the executive and the legislative branch, is discussed by Suleiman, op. cit., pp. 167-69.

64. For a discussion of this point, see, for instance, Ørvik et al., op. cit., p. 132, and George, op. cit.

65. As mentioned previously, such a form is more consistent with the limitations of search procedures. The organization may thus influence the minister's effective choice by limiting the actual search initiated prior to the presentation of the case. This point is given careful consideration by George, op. cit.

66. For a thorough discussion of the merits of institutionalizing this procedure, as compared with the preparation of briefs by a central staff after the fashion of the NSC under President Nixon, see George, op. cit., and the ensuing debate with Destler, American Political Science Review, 66 (1972), pp. 786-95.

67. This is the method used, with much success, by Suleiman (op. cit.) and Bacchus (op. cit.).