Pergamon Accounting, Organizations and Society, Vol. 23. No 4. pp. 41 l-434. 1998

( 1998 Elsevier Science Ltd All rights rrsavrd Printed in Great Britam

0361.36B2/98 $lO.OO+O 00

PII: SO361-3682(97)00018-4

RELATE MANAGEMENT ACCOUNTING RESEARCH TO MANAGERIAL WORK!*

S. JtiNSSON University of Gotbenburg

Abstract

This article argues that management accounting research should be more aligned with managerial work. Managers work with words. This means that managerial conversation and the use of accounting informa- tion in such conversation should be studied. Methodology is developing rapidly in other areas of social science where a linguistic turn has made an impact. Managerial work, characterised by brevity, variety, and fragmentation, provides the context in which accounting information is used. Management is thus described as a ccFoperative game where communication is central to attention direction as well as prob- lem solving. An illustration is given of the kind of studies deemed necessary for management accounting research to progress towards managerial re!evance. ,(;: 1998 Elsevier Science Ltd. All rights reserved.

It seems like management accounting research has been out of focus for some time now. Various reasons can be named. There is clearly a lack of empirical input to the theorising which is increasingly drawn towards clean ontological assumptions suitable for rigorous analysis. Empirics tend to be limited to quick survey studies which fit into the publication requirements of the main stream. There are also some complaints of a lack of relevance (Johnson St Kaplan, 1987; Cooper, 1990) of standard cost accounting technologies for complex real-world settings. Case study approaches have been recommended (Ton&ins & Groves, 1983; Hopwood, 1983, 1987; Scapens 1990). Researchers and educators alike have been exhorted to get in touch with the real world. Still no breakthrough has been claimed.

been presented (Dent, 1991; Bhimani, 1993). Studies where managerial action to adapt the organization to changing environments have been studied. At best these studies provide ontological insights by using the history of the organisation as the ground against which the figure of current change is depicted. They may at times also be connected to current debates and fashions of accounting technology.

Concern has been expressed for the limita- tions of contingency studies relying on cross- sectional methods and statistical inference (Chua, 1986; Merchant, 1985). As a counter point to such studies longitudinal ones have

The most promising aspect of these studies is, however, that they make managerial action visible in accounting studies. Management action seems oriented towards change rather than choice; lean production, customised ser- vice, multi-domestic strategies provide interest- ing challenges to management research. In this area the management of strategic change, no doubt stimulated by the celebrated success of Japanese companies on western markets, has been a dynamic field. Community and network have been added to the traditional dichotomy of Market and Hierarchy. Virtues like loyalty and trust stand as ontological alternatives to

The research on which this article is based was financed by a grant (90-354 03) by the Bank of Sweden Tercentenary Foundation. Final revisions were carried out while the author held an Erskine Fellowship at the University of Canterbury. (Christchurch.

411

412 S. JiiNSSON

opportunism and individual maximisation of utility. The study of organizational learning has advanced and changed our views of the role of communication (Hedberg & Jonsson, 1978). We realise that communication is not only a matter of information technology but a crucial factor in the social construction of practices. A further complicating aspect of the modern cor- poration is the differentiation of technology and their attached knowledge communities (Boland & Tenkasi, 1995). Managers have to organise and maintain knowledge intensive communica- tion around the redesign of products as well as processes in cooperation between specialists (in house and outside). There is likely to be misunderstanding which cannot be easily resolved by resorting to power or hierarchy. Communication through several channels and sensemaking (weick, 1995) will be a major occupation.

All these aspects of the management task effect managerial work which, in turn, will influence the selection and use of management information. The relevance of accounting infor- mation will depend upon its ability to relate to the work managers do.

This paper attempts to introduce such a managerial perspective for the development of management accounting research. A perspec- tive which would align it with current develop- ments in management and social research focusing on the communicative aspects of managerial work. First the reader will be reminded of some of the results of empirical studies of managerial work. Then the meaning of an approach to management accounting from the managerial side is discussed. The dimensions thus introduced will support an illustrative example of how communication in a management team meeting can be analysed. The paper aims to stimulate readers to explore the area further, to persuade rather than prove.

MANAGERIAL WORK

The first systematic, empirical study of man- agerial work was probably done by Carlson

(1951). Mintzberg (1973) was inspired by Carlson but applied a more intensive observa- tion method (participant observation) than Carlson who used a diary method in combina- tion with interviews. Mintzberg observed his managers for five days each while Carlson col- lected data on one month for each manager. Kotter (1982) used a combination of inter- views, questionnaires, and participant observa- tion over two periods for each participant. These three studies will be used as indicators.

Carlson structured his observations in inci- dents for which the following data were recor- ded; where was the work done, what time did it take, who participated, what manner of com- munication was used, what was the nature of the business, and what kind of action did it imply (decision, inquiry, command, etc.).

When reporting the results, Carlson com- plained that the top managers spent so much time away from the company for non-commer- cial contacts. When they were in, they spent too much time in inefficient meetings. Meetings took too long and were not properly co-ordina- ted with other forms of control. Far too little time was devoted to analysis and planning. Why not use subordinates in all these meetings! They could better devote their full attention to the matters dealt with by the committees and working parties! Managers also seemed system- atically to overestimate the time they spent managing-by-walking-around.

Carlson found that the average time alone in the office before being interrupted again was 8 minutes. A typical top manager (CEO) worked undisturbed in his office for more than 28 min- utes on only nine occasions over a period of 35 days.

Contact diagrams showed that functional business dominated the agenda. On the average about 15% of the time was spent on corporate matters and the rest was spent with represen- tatives of functional areas. Coordination by res- ponding to one functional area on the basis of current knowledge of the other areas seems to have been a critical task for Carlsons managers. Only 6.3% of the incidents were classified as ending in a decision. The most frequent activity

RELATE MANAGEMENT ACCOUNTING RESEARCH TO MANAGERIAL WORK! 41.3

was getting information (39.6%) followed by give advice, explain (14.6%).

Mintzberg (1973) worked with a chronology, postal protocol and a contact protocol that was coded for a week for each of the observed managers. Mintzberg summarises his observa- tions on the nature of managerial work in terms like relentless pace and brevity, variety and fragmentation. It seems like managers prefer activities that are current, specific and well defined before the abstract and non-routine ones. They work verbally with fresh informa- tion rather than analytically with systematic information.

Mintzbergs study is probably most well known for its typology of leader roles; rela- tional, informational and decisional roles. The empirical basis for these role descriptions must, however, be considered rather thin. If one con- siders the fact that roles are developed over time and in interaction with others the one week observation period is not sufficient. Still the criterion question in Mintzbergs classifica- tion of activities into roles: Why did the man- ager do this? indicates the multitude of tasks managers are obliged to carry out. Mintzberg refers to Katz and Kahn (1966) who state that leadership is related to incompleteness and uncertainties inside and outside the organisa- tion, which means that management could be described as an unprogrammed activity. Bar- nard (1966) claims that the most important task for the manager is to maintain organizational activities. This adds up to a description of man- agements main task as being to assure that the organisation is effective, stable and has an abiI- ity to adapt. This requires continuous vigilance concerning environmental, task, person, and situation variables. Emergencies are normal (Walker et aZ., 1956, p. 76). Nonetheless there are routines and recipes for certain types of problems (acquisitions, turn-rounds). Mintzberg discusses the possibilities of programming these recipes and whether this could not be done in co-operation with analysts who could distance themselves better than the manager. Mintzberg is thinking about academics here, but the task of extracting patterns in managerial

behaviour and programming them could as well be done by staff departments. When he gives advice on how the situation could be improved it points in the direction of more team work in top management (which, obviously, implies communication).

One cannot help wondering if, perhaps, all these intelligent, successful managers indulge in managerial work characterised by brevity, variety and fragmentation because it is an effi- cient way of running a company?!

Kotter (1982) introduces managerial work in terms of its complexity. It is not so much the analytical task that is difficult but the intensit) of the competitive pressure which provides for variety and change. Dilemmas of complexity are illustrated; setting goals in spite of the fact that basic underlying conditions are uncertain, balancing short term return and long term development, keeping informed, criticising and motivating subordinates. These are issues with- out analytical solutions.

Kotter finds that the managers he studied were not analytically inclined, neither are they generalists. Managers are socially talented and they have usually acquired an extensive knowledge of an industry in their career to be able to be accurate in intuitive judgement. In this sense they are specialists.

Kotter found that managers took on their tasks in a similar manner. Early-up to six months or a year-after their appointment they devoted their attention to the establish- ment of an agenda and building the necessar) network to carry it through. Such an agenda coincided only partly with the official business plan of the organization and it had content as well as process aspects. During this initial per- iod the new manager aggressively sought information, primarily from those he already had trust relations with, by asking questions using his own experiences to guide the direc- tion of the questions. The sequence of meetings and discussions with co-workers are often unplanned and disjointed. Choices are made on intuitive grounds to achieve specific programs or projects, that can be expected to contribute to several goals at the same time, and for which

414 S. JiiNSSON

a mandate to implement is sought. This goes on continuously, bit-by-bit in a time consuming manner during the first phase. The content of the agenda is characterised by loosely coupled goals and plans based on business ideas with mixed time horizons and covering a broad spec- trum of issues. They were not put in writing.

Building networks is time consuming. It means spending time with others, and managers devoted considerable time to building relations assumed beneficial to the implementation of the agenda. Relations often reach far beyond the immediate coworkers and across formal organizational boundaries. The manager in his (or her) role as boss has an interest in promot- ing teamwork and eliminating politics in his immediate neighbourhood. On the other hand there is a need to build networks which can be used to short-circuit the formal organisation to promote implementation of the agenda.

Kotter attributes differences in the manner in which all this is done to background and career experience. The more comprehensive the experience of the industry the less time was devoted to agenda- and network-building and more to implementation. Such differences in the allocation of time to networking have also been documented by Luthans et al. (1988). Here focus is on successful (fast career) and effective (good responsibility centre perfor- mance) managers. While successful managers spent 48% of their time networking, effective managers spent only 11% of their time on this kind of activity.

Once the agenda and network were estab- lished they asserted a more direct influence on the daily work of the manager. The agenda directed attention to its issues and the network framed how they were approached. In this way managerial agendas affect significantly beha- vioural patterns of the company. On the issue of whether it is possible to detect differences between more or less effective managers, Kotter claims that even the small sample that he studied gives reason to conclude that several completely different behavioural patterns, see- mingly based in contingency-specific solutions, seem effective.

These empirical studies of managerial work have some notable commonalities: (1) Manage- rial work is mostly done in groups and in face- to-face communication; (2) Decisions are rare events-asking questions and getting informa- tion are the most frequent activities; (3) Man- agers work with fresh, verbal information in a fashion characterised by brevity, variety and fragmentation in a stream of events where agendas related to but separate from the formal plans of the company are at work. Managers want to shape and change the organization to something better, and they do it with words. To achieve this they build networks of personal relations. They may be assumed to behave this way not because they are stupid and ignorant but because it works.

THE EMBEDDEDNESS OF MANAGEMENT ACCOUNTING

Management accounting, seen as an indicator of structures of responsibility, is embedded in a modernist frame, the specificities of which may vary somewhat between countries (Garner, 1954; Loft, 1986; Jonsson, 1991, 1996). The universal modernist frame of reliance on con- trol from the top, use of scientific knowledge to improve the world, etc., is moderated by parti- cular experiences of war, paths of internationa- lisation, values as to the relation between individual and society, institutional histories, etc. After the oil crises, deregulation, European integration, and the liberation of trade the modernist frame is challenged. Regulation, which was largely carried out in a national set- ting, has been replaced by a view of harmoni- sation across several cultures. The modernist conception of rationality based on a given and accepted set of values is replaced by accommo- dation to several sets of values. This is espe- cially true for small countries with open economies. Reformers of management refer to multiple sets of values, often abstracted and labelled beyond comprehension (customer satisfaction, value-added, multidomestic com- petition).

RELATE MANAGEMENT ACCOIJNTING RESEARCH TO MANAGERIAL WORK! 415

The last decades have seen a management reform movement argued on the basis of ideas linked to lean production (inspired by Japanese models) and customer focus (inspired by American ideas of value added).

Management accounting, which used to be securely embedded in a context of large scale production of a limited number of products, lost its bearings in this new situation of global strategy turmoil. Criteria for what constitutes relevant information have become ambiguous and uncertain. Relevance may have been lost. Still, product costing has hardly changed in its basic structures (Robinson, 1990; Ask et al.. 1996) even if the ABC approach, with its analy- tical and decision making focus, caught a lot of attention. To regain contextual sensibility man- agement accounting needs to realign with managerial work. Rather than sticking to a score card function (Simon et al., 1954) it should support managers when they make sense of managerial situations, direct attention, and solve problems. No longer will researching the everyday accountant (Tomkins & Groves, 1983) suffice. It is the everyday manager who will determine what is relevant, the everyday manager in context. The search for a new alignment therefore should be clinical. For results to be useful, management accounting research should be tuned to what goes on at the clinic and focus on how management accounting becomes embedded in new social or managerial structures. If this is accepted as a reasonable point of departure, the formidable task of maintaining the systematic and concep- tual basis while developing the communicative aspects of management accounting lies before us.

WHAT WOULD IT MEAN TO APPROACH MANAGEMENT ACCOUNTING FROM THE

MANAGEMENT SIDE?

If a more management orientation to man- agement accounting were to be adopted, we would not want to diminish the usefulness of analytical applications of management account- ing techniques in any way. But we would be

interested in how management accounting information is used in managerial discourse, i.e. when attention is directed and problems solved (Simon et al., 1954). These latter activities will be based on soft information in terms of intui- tive definitions of situations where the joint social construction of action premises goes on. The analytical use of management accounting information is likely to be referred to staff units for confirmation of hunches or the calculation of aspects of alternative lines of action. Line managers, working in real time in management situations characterised by brevity, variation and fragmentation, are likely to use account- ing information in conversations with others, be it joint problem solving, implementation processes or attention direction, to establish rules of the game, commitment to tasks and policies, and for testing half-baked ideas. Such conversations are usually cooperative games rather than competitive ones. This means that the argument of one participant will build on and add to that of another participant and the joint product will be beneficial to both. It also means that the game is not specified in advance but generates its own contextual meanings. Managerial conversations do not only convey information or solve problems but also serve to generate symbolic capital and trust. Participants build identities as competent managers by rele- vant contributions which are presented in accordance with how things are done in the given context. This effectively discourages opportunism, since participants will meet again in new conversations, and because performance in conversation will be evaluated by peers. This poses problems for an outside observer. Help from participants is required in sorting out the implications of what was actually said, meant. and understood during the conversation.

MANAGEMENT AS A CO-OPERATIVE GAME

In order to be able to analyse managerial conversations as cooperative games a set of concepts and conventions for relating them are needed. A basic ontological assumption is that

416 S. JiiNSSON

participants work with each other and that a game is finished by mutual agreement rather than by applying some criterion of winning. Conversations in themselves have some sys- temic and ritualistic constraints which have to be met for them to function as a medium of interaction. Furthermore we are interested in implications for managerial action, not only in the conversation itself.

The constitution of conversation Goffman (1981, p. 14) specifies the following

systemic requirements for a meaningful con- versation:

1. A two-way capability for transceiving acoustically adequate and readily interpre- table messages.

2. Back-channel feedback capabilities for informing on reception while it is occur- ring.

3. Contact signals: means of announcing the seeking of a channelled connection, means of ratifying that the sought-for channel is now open, means of closing off a theretofore open channel. Included here are identification and authentication signs.

4. Turnover signals: means to indicate the ending of a message and the taking over of the sending role by next speaker@ the case of talk with more than two persons, next speaker selection signals, whether speaker selects or self-select types).

5. Preemption signals: means of inducing a rerun, holding of channel requests, inter- rupting talker in progress.

6. Framing capabilities: cues for distinguish- ing special readings to apply across strips of bracketed communication, recasting otherwise conventional sense, as in mak- ing ironic asides, quoting another, joking and so forth; and hearer signals that the resulting transformation has been followed.

7. Norms of obliging respondents to reply honestly with whatever they know that is relevant and no more.

8. Nonparticipant constraints regarding eavesdropping, competing noise, and blocking pathways for eye-to-eye signals.

Inside such a framework there are numerous ritual constraints at work to regulate how par- ticipants should act to preserve everyones face, express disapproval, sympathy, greetings etc., and also how potentially offensive acts can be remedied, and offended parties can induce remedy. There is also a specific ordering device, adjacent pairs, which caters for the flow of conversation by statement-response links. The variety within such a framework is overwhelm- ing and the micro-analysis of discourse through conversation analysis has to deal with them. Here is not the place to go deeper into these matters. It is however necessary to indicate an approach to how participants work out impli- cations of what is said in conversation, i.e. how sensemaking (Weick, 1995) is accomplished. For this purpose Grices Cooperative Principle (1989, originally presented in 1975) will be used.

Implicature As already mentioned, conversations are

typically co-operative efforts and could be thought of as quasi-contractual in the sense that there is a common immediate aim. Contribu- tions are mutually dependent, and the interac- tion should continue until all or both parties agree to terminate. The general formulation of the cooperative principle is : Make your con- versational contribution such as is required at the stage at which it occurs, by the accepted purpose or direction of the talk exchange in which you are engaged (Grice, 1989, p. 26). Grice distinguishes four categories under which fall maxims to be followed to live up to the cooperative principle. These categories are:

l Quantity: make your contribution as informative as is required and no more!

l Quality: try to make your contribution one that is true!

l Relation: be relevant! l Manner: be perspicuous!

RELATE MANAGEMENT ACCOUNTING RESEARCH TO MANAGERIAL WORK! 417

Conversational implicature can be achieved by failing to fuffil the maxims (violate a maxim, opt out, clash between maxims, flouting) under these categories. A hearer will repair the failure by finding an implication that wiIl restore the cooperative principle, i.e. make sense of the implication. (He said that p; he is following the maxims of the co-operative principle; he could not have said that unless he thought that q; he knows that I will recognise that he thinks that q is required; so he has implicated that q.).

Example: A: How is X getting on in his new job? B: Oh, quite well, I think; he likes his col-

leagues and he has not been to prison yet. (Grice, 1989, p. 24)

Here B fails to fulfil the maxims of Quantity by providing more information than requested. A will have to work out the implications.

One of the points with Grices Cooperative Principle is that it gives illustration, in an intui- tively satisfying way, to the work participants in conversations do to bridge gaps and rectify fail- ures in conversational cooperation. To accom- plish something similar the analyst will require context information as well as participants help in arriving at conclusions. Research into conversational implicature must therefore work with data collection on at least two levels, the conversation itself and the context relevant for uncovering the implications.

Figure 1 illustrates a conception of communi- cation in two dimensions. We have now dealt, admittedly briefly, with the horizontal axis.

The content of the message is given meaning when placed in context. This is done by sender and receiver, who may have different experien-

Sender

Content +

Context

Receiver Fig. 1. The factors of communication.

tial backgrounds. Complexity arises because of these differences. Not only are experiences themselves different, but they may have gener- ated different kinds of structural and institutional frames for sensemaking. Institutions like trust and commitment, roles and teams can serve as reducers of complexity, because they enhance the use of codes (Bruners (1990) paradigmatic cognition.). However, communication across institutional boundaries may require extra efforts to translate the relevant institutional contexts. The next step is to consider the institution building aspects of managerial conversation.

BIJILDING INSTITIJTIONS

Complexity arises from the fact that informa- tion emerges as the receiver interprets the message. The sender may have an intended meaning with the message, e.g. to arouse a cer- tam pattern of behaviour in the receiver. But the extent to which that intention is realised is largely dependent upon what meaning the receiver ascribes to the message (if the message is at all noticed in the noise of everyday urgen- cies). This emergent view of information (Moore & Carling, 1982) is crucial for the ana- lysis if narratives constitute a significant part of the conversation. If sender and receiver have developed mutually understood codes in earlier interaction they may have little difficulty in reaching similar or the same interpretation of the content of the message. In this paradigmatic mode (Bruner. 1990) the content of the mes- sage is a matter of coding and decoding (plus channel capacity). Information may be said to be contained in the message (Moore & Carling, l982)- the container view. This case is, however, of less interest in a managerial per- spective because it illustrates programmed control (Simons, 1990) where codes are used in a more or less technical communication. More interesting here are situations where members of managerial teams are assumed to have different experiential bases for their interpretive work or where there is communi- cation between teams. Then it is likely that the

418 S. JiiNSSON

same message is attributed different content by different participants. Variability in interpreta- tion must be assumed to be the normal situa- tion in communication between communities of knowing (Boland & Tenkasi, 1995).

The fact that sincere interpreters can arrive at different understandings of the same message brings out the issue of truth. How can it be determined that the story contained in the message is true? Most of the time we cannot be sure. We have to rely on trust. Even if it could be determined that the story was untrue it could be persuasive. The story could have internal consistency and reveal a connection between variables that is interesting, or it may provide the basis for a useful concept. Such generation of relations, concepts, or models of interpretation which seem useful in under- standing other situations have a value. They contribute to the language of organizational interpretation. Such stories are instrumental in generating language which extends vocabu- laries. The meaningfulness of them consists of their making sense. The truth concept of pragmatism (lames, 1974; Mead, 1934) comes to mind. If it works in your experience, if you are ready to act on this new insight after having been frustrated, the information has truth value in a pragmatic sense. Provisional truths may be tested through conversation (Do you see what I see?) and through action. If action fails data will be reconsidered, but if it succeeds, this experiential confirmation will constitute learn- ing and the portfolio of practices or the inter- pretive repertoire (Gilbert & Mulkay, 1984; Potter & Wetherell, 1987) may expand. Repe- ated successful action will prime the inter- pretive patterns (Wilson, 1980) of actors and teams so that they will be more likely to be evoked in similar situations. In time custom and even culture are formed.

BUILDING BLOCKS OF MANAGERIAL INSTITUTIONS

A significant difference between managerial behaviour and the behaviour assumed in markets

is that managers wiIl meet their partners in dis- course repeatedly and thus have an interest in building relations which will facilitate future communication. Such effects will emerge if the manager behaves consistently. Other members of the organisation will then be able to see the implications of communicated messages.

Rule -following According to Wittgenstein (1953) and Winch

(1958) meaningful behaviour must meet two criteria:

1. the actor must be able to give a reason (account) for an action, and

2. the actor must be bound by the act now to do something consistent with that act in the future.

When these two criteria are met the actors behaviour may be said to have a sense.

The first criterion means that the actor must be capable of providing a justification or an account for the act. The actor will usually not have to specify all the alternative acts that were at hand, but it is obviously necessary to show that there was a choice (at least to avoid action) and the reason for the actual choice. (If there was no choice there is no initiating actor and no responsibility.) The reason does not have to be true but it must be recognisable as a reason and it must not be incompatible with other information about the situation.

The second criterion establishes the rule-fol- lowing or commitment aspect of action. As the actor makes public the reason for the act he or she also announces an intention to behave accordingly in the future. A principle has been applied that is elevated to be a rule in the repertoire of rules that generates the beha- vioural pattern that in turn constitutes identity. Hereby accountability is established. The actor declares a willingness to be held accountable for this and consequent acts. This giving of rea- sons for acts constitutes explanation to the environment. As such, it has to be geared to the conceptual framework (paradigmatic meaning) or constitute a self-contained story (narrative meaning). The environment will signal its

RELATE MANAGEMENT ACCOUNTING RESEARCH TO MANAGERIAL WORK! 419

expectancy of an explanation by holding the actor responsible for this and future acts. Com- petent rule-following will lead to increased trust and a granting of a status of self-manage- ment, i.e. to be allowed to handle sensitive matters and warranty relations without inter- ference from other members of the group. This is akin to being considered Mundig (in German), which means having the right to sign contracts, and being bound by ones word.

Rules of the game When a person understands the reasons for

the acts of others and expects them to be bound by their commitment to behave consis- tently in the future he or she may act confi- dently and effectively because of the trust in the rules of the game. Garfinkel (1963) illus- trates a conception of trust via a description of games. The basic rules of a game involve fram- ing a set of moves which the player has to choose from, which the player expects to be binding for him as well as other players, with the player expecting other players to expect the same thing in return. This gives rise to the constitutive expectancies of the game. Trust is then defined as constituting a situation where individuals in their interpersonal relations are governed by constitutive expectancies.

When trust is breached, e.g. by actor moves that violate the rules of the game, Garfinkel (I 963) found that individuals engaged in vigor- ous efforts to normalise the situation, i.e. return it to the original constituent expectan- cies. He also found that persons who tried to retain the original rules of the game were more frustrated in their normalising efforts than per- sons who were willing to redefine the game. This indicates that the rules of the game have strong normative power. Wieder (1974) extrac- ted maxims from interviews with inmates in a half-way house for reformed drug addicts that seemed to guide behaviour. He could then demonstrate how these maxims were used to regulate conversation. One of the maxims, in fact the primary one, was Above all do not snitch. When Wieder had a friendly conversa- tion with an inmate it was sometimes abruptly

stopped by the inmate saying you know 1 wont snitch. Analysing what this utterance achieved, Wieder found that it (1) defined what Weider had just said as a request to snitch which allowed the inmate not to answer. It would have been illegitimate for an insider to answer. It also (2) served to confirm that Wieder was an outsider and not a friend. If Wieder would have continued his line of ques- tioning he would risk being seen as an incom- petent, as someone who does not understand the Code.

Roles of the game When an actor meets constituent expectan-

cies and engages in meaningful behaviour, every act, consistent with the expectations, serves as confirmation of a role in the game. Consistency in behaviour provides a stable basis for other actors rational calculation and efficient response and binds the actor further to role consistency.

In sociology there has long been a clash between two approaches to role theory, struc- turalist and interactionist. Structuralists see the role concept as a set of social norms constitut- ing a part of the culture that influences actors in a given situation (Turner, 1985). Society in principle is a system of social norms and beha- viour is an expression of how well the indivi- dual has intemalised and adapted to these norms, Critics of this view ask how were these norms established in the first place and how did they achieve a status independent of the role?

The symbolic-interactive view sees the role concept in more of a communicative light. The role has a character of a relation and the indivi- dual assumes and forms the role in interaction with the environment. The communicative capacity of humans makes it possible for them to see a situation not only from their own per- spective-not even only in alters perspec- tive-but also from a third position and see the interaction between ego and alter from the outside, objectively as it were. This interac- tive ability rends behaviour meaningful, i.e. the ability to foresee the reactions of others to own behaviour. Role expectations are mutual and

420 S. JiiNSSON

situation specific. The role is formed in interac- tion on specific problems.

The fundamental difference between these two views-the role given by the expectations of the environment (with role conflicts if the individual is unable to meet them) on the one side and the role emerging and changing in communicative interaction in a situation on the other side-may not be as fundamental as it once seemed. Turner (1985) sees convergence between the views over the last decade. The differences seem to have been inspired by the different modes of data collection and analysis that the two schools have practised. In both cases it seems reasonable to assume that rule following is a generator of roles. When people assume roles they certainly put limitations on their repertoire of behaviour, but they also become more predictable and they may reduce complexity (Wildavsky, 1975).

Teams A set of mutually constituted roles is a team.

The members of a team establish a division of labour and constituent expectancies in an inner dialogue (Ionsson, 1992) in which a unit is chiefly concerned with translation from narra- tive meanings to paradigmatic ones, i.e. from stories to principles. Bruner (1990) maintains that there are two modes of cognition, the paradigmatic and the narrative. The paradig- matic mode starts from definitions, models and deductive logic and provides a rational analysis inside a problem frame. Data become meaning- ful by being processed through a model. A technology is applied to the data. The narra- tive mode builds on our capacity to make sense of our experience by narrativising it, tell a story, put it into script as it were. The narrative gets its meaning from the internal coniigura- tion. The narrative is good and worth remem- bering if it is exciting, plausible and persuasive. Paradigmatic information is acceptable if it is logical, noncontradictory and consistent. The narrative is social and it is through narrative that we build and maintain culture and praxis. When the narrative is established well enough to be taken for granted it has been conceptua-

lised into the paradigm or integrated in back- ground context.

This translation work gives members of the group opportunities to build trust and profes- sional competence in experiential learning cycles (Kolb, 1984) and it builds on the giving of accounts. A competent person must be able to account for what happened. It is part of the definition of competence. You should always know what you are doing and why. If you are competent then you are responsible for the consequences of your actions.

Everyday life is built around giving accounts (Garfinkel, 1967). The network of social life is constituted and maintained by story-telling, or accounts (explanations) of things that seemed to have no meaning before. Once we are given the account that bridges between that which we already take for granted and that which seemed odd we can extend the area of normal- ity yet another inch. Inside the constituting network of the social we can then use the paradigmatic for effective communication in coded messages. Management accounting is such a paradigmatic code used by responsible, competent persons who manage resources and personnel over which they are granted jurisdic- tion by others. The technical has its own rules of communication. By applying them properly membership in the class of responsible team members can be maintained (Munro, 1994). When there is error, when there is a variation from budget or some other expected perfor- mance target, an explanation is in order. Typi- cally we use a story of some kind. If we say It was due to the fluctuating currency situation that may be an acceptable explanation, but if we say the deviation from budget is due to my wife being seriously ill (I cannot concentrate on my job) it will probably not be accepted as a legitimate explanation. At best we might get away with Dont let it happen again! (Which means that we lose part of the trust that was put in us.) The manager receiving our explana- tion may feel sympathy with our problem, but the role of principal does not allow exceptions from the rules of the game. Membership puts limits to what kind of narratives we may

RELATE MANAGEMENT ACCOUNTING RESEARCH TO MANAGERIAL WORK! 421

allow ourselves as explanations to error (Munro, 1994).

Membership work thus amounts to giving accounts for our activities and developing rou- tines for doing so. In this way routines become the constituents of social order (Giddens, 1984). But routines are also there to keep things apart. Not to mix up company time with personal time, company money with private money and admissible accounts with inadmissi- ble ones. A constant labour of division goes on and part of membership work is to keep a clear distinction between what is in and what is out. This requires attention to the rules of the game by members: The activities whereby members produce and manage settings of orga- nised everyday affairs are identical with mem- bers procedures for making those settings account-able (Gartinkel, 1967, p. 1.).

Between teams-discourse Communication between teams is complica-

ted. Interaction between teams is less frequent than in the inner dialogue of teams and it is more abstract in the sense that it rarely takes place in the presence of the work process. Communication typically takes place in a meet- ing between representatives of the involved teams. If teams develop special expertise, which might be the main purpose of a decen- tralisation but also a consequence of evermore differentiated technology, they may also become intolerant towards information, advice, orders, intervention from other groups. If teams have developed a role in a large unit it will be expected to behave consistently with that role and may need to negotiate contracts to change that role. It is also likely that the same data will be interpreted differently by different communities of knowing due to the different experiential basis for the interpretation. Then narratives will be used in reaching an under- standing. But a precondition for reaching an understanding is that the conversation develops as a co-operative game.

There are two, possibly alternative, approa- ches to the tools people use to make sense of the world which are of interest here; Moscovicis

theory of social representations (Moscovici, 1984; Moscovici & Doise, 1994) and the con- ception of interpretive repertoires (Potter & Wetherell, 1987; Gilbert & Mulkay, 1984).

Social representations are mental schemata with abstract as well as concrete elements- concepts and images. In most representations concrete images are most important in which case the representation is built up around a figurative nucleus. These social representa- tions are assumed to underpin attributions or the causal explanations people give for events (Potter & Wetherell, 1987, p. 140). The repre- sentations are social because they originate in the course of social interaction and because they provide an agreed code for communica- tion. Moscovici also claims that social repre- sentations are social because they serve to homogenise and unify groups. Since they pro vide a conventional code of communication and are used in the sense-making process, all who share a representation will agree in their sense making. The members of a group share representations and therefore representations can be used to identify a group. (This is a pro- blematic statement that we will come back to.)

Social representations have a constructive effect. When a group makes sense of the world, it is constructed in terms of social representa- tions. Two mechanisms are proposed for how people cope with new experiences-anchoring and objectification. First, anchoring is used to assign the new object to an element or category in an earlier representation, then the new object is transformed into a concrete, pictorial element of the representation to which it is anchored, and the new version of the repre- sentation is diffused in conversation.

Potter and Wetherell (1987, pp. 142 ff.) challenge this theory of social representation (referring to their own research and that of col- leagues) and the thrust of the critique is directed towards the assumption that the community of representations establishes group identity. This assumption is problematic becauseit creates a vicious circle of identifying groups through representations and representations through groups. There are also empirical problems

422 S. JiiNSSON

connected with the use of representations in the sense that a representation could be used in explanations or merely mentioned (the left wing press have claimed that the riot was caused by poor housing, but...). It is also pos- sible to distinguish between use in theory (general formulation) and use in practical situa- tions. Finally there is the issue of social repre- sentations being cognitive or linguistic. Moscovici seems to view them as cognitive phenomena. Researchers are then faced with the problem of constructing a neutral record on the basis of secondary phenomena. How could one recognise an instance of anchoring in the interview transcript?

To remedy some of these problems Potter and Wetherell offer the notion of interpretive repertoire which has been developed in colla- boration with several colleagues in empirical discourse analysis. The point of their offer is that they focus on how members cope with accounts in interaction-discourse in con- text-rather than on cognitive aspects-what lies behind. A study by Gilbert and Mulkay (1984) is used to illustrate this phenomenon.

After identifying a group of biochemists in the U.K. and U.S.A. who worked in the same area Gilbert and Mulkay collected publications of respondents and conducted interviews where accounts of biochemical research work were elicited. There were two competing the- ories in this prestigious area and adherents to both were represented in the network of 34 of the most productive biochemists. But there were also two modes of accounting for scienti- fic work. The formal one of scientific writing, and the informal one of interviews, seemed to generate different accounts of the same action. In a scientific article it was the results (of standard experimental work) that suggested the model, but in the interview it was the model (a dramatic revelation) which suggested the results. On the basis of numerous differen- ces in accounts Gilbert and Mulkay arrived at the conclusion that there exist two broad inter- pretive repertoires that explain the differences in accounts. These were labelled the empiri- cist and contingent repertoires, respec-

tively. Interpretive repertoires are recurrently used systems of terms used for characterising and evaluating actions, events and phenomena (Potter & Wetherell 1987, p. 149). The basic principle in the empiricist repertoire is that actions and beliefs are a neutral medium through which empirical phenomena make themselves felt. The scientist is forced by the observations and professional rules to take actions (accept a hypothesis). In the contingent repertoire, the basic principle is that actions and beliefs are crucially infhtenced by factors other than the empirical phenomena. The sci- entists used both. They used the empiricist repertoire to warrant their beliefs by referring to how experimental evidence and correct application of the rules of scientific work pre- vented them from having any other beliefs. Given the obligations of the empiricist reper- toire scientists can make no mistakes. When an error occurs scientists use the contingent repertoire to explain. The contents of the con- tingent repertoire were more varied than the empiricist one. Subjective elements were inclu- ded like prejudice, narrow disciplinary per- spective, threat to status, lacking experimental skill, personal rivalry etc. When all the accounts of these researchers were combined virtually everyone in the network had their work explained away by at least one other scientist. Gilbert and Mulkay found that researchers used asymmetrical patterns in explaining right and wrong beliefs. The empiricist repertoire- facts arise naturally from empirical findings- was used to explain right beliefs (other sci- entists regularly seem to get them wrong), and the contingent repertoire-provided a package of non-scientific resources for explaining error (Potter & Wetherell, 1987, pp. 152f.).

Scientists manage to live with this asymmetry in accounts of scientific work by using the truth-will-out-device. Even if the progress of science is sometimes retarded by personality problems and power struggles, the truth will win in the end through experimental evidence. This rhetorical device serves to reconcile potential contradictions between the two repertoires, but it also re-establishes the primacy

RELATE MANAGEMENT ACCOlNTING RESEAR(:H TO MANAGERIAL WORK! 42.3

of the empiricist repertoire, which may be its main contribution.

Potter and Wetherell seem to perceive these two approaches as exclusive alternatives, but surely they are quite compatible. Moscovicis main thesis is that of group polarisation (Moscovici & Doise, 1994), which means that collective decisions in groups tend toward radi- cal positions and members retain the groups position after such an occasion of consensus building. If, however, restrictions are laid on the groups discussion, for instance in the form of an agenda or a time schedule, the group will tend towards compromise and group members will not retain the opinion expressed by the groups compromise decision. If this aspect is added to the account of the social representa- tion approach given above, compatibility may be posited. The ontology related to social representation will be valid for the inner dialogue of teams and that related to interpre- tive repertoires will be valid for teams outer dialogues. Of course, there is a need for teams to have a contingent repertoire to make sense of unexpected situations, but the social insti- tutions of the team will build on common representations. When, however, the team communicates with other teams there is an obvious need for more than one interpretive repertoire to make sense of the actions of the other team. Such translational vocabularies may be primitive and based on stereotypes initially, but they will develop as interaction with the other team develops. Some terms may find their way into the teams own repertoire.

Clearly, it has been demonstrated that also the vertical relation in Fig. 1 (between Sender and Receiver) is a complex one. The next pro- blem is to probe into how communication in managerial work can be observed and analysed.

THE EMPIRICAL BASIS FOR THE ANALYSIS OF COMMUNICATION IN MANAGERIAL

WORK

Conversational analysis is often criticised for being merely interpretive and not replicable.

Alternative interpretations of the same text are possible and therefore interpretations are subjective. Counter arguments are that the con- versational analyst does not simply interpret or read the text, but examines participants actions and achievements in context and through the structure of the conversation. Furthermore the conversational analyst works with recorded facts (the recorded conversation did in fact take place, and may be assumed to be sincere since in the company of peers it is not possible to uphold play- acting for any period of time). The empirical evi- dence is available for others to analyse to a larger extent than most other methods. The conversa- tional analyst does not refer to speculative assumptions about mental states (like attitudes, rational expectations, preference orders, or utiles) in drdwing his or her conclusions. Buttny (1993) develops a conversation analytic constructionist perspective which seems suitable for further development of tools for analysing managerial conversation and how it accomplishes changes in pracrices. Drawing an analogy between the role of the tape recorder in the study of social interaction to that of the microscope in biology, he points out the opportunity to reveal features which are commonly seen but unnoticed during actual con- versation through repeated observation of sequences (pp. 56f).

In the following section an illustrative case is presented of how analysis of conversation in context could be used to gain insight into the development of accountability and manage- ment accounting in an organisation.

SETTING RIJLES FOR INTERNAL RENTS

The conversation about to be analysed is a sequence in the early part of a videotaped meeting of the top management group of one of the down town municipal districts in one of the major cities in Sweden. The background to the study was that the City had decided to reform its organizational structure in 1989. In the reform decision was included a contract with the IJniversity to evaluate the reform over a five-year period. This gave rise to a very large,

424 S. JBNSSON

multidisciplinary research project from 1990 onwards.

The new organisation meant that the city, which was previously organised in large func- tional departments (Social, School, Culture, Lei- sure were involved in the reform), was now divided into 21 districts, each with their own political executive committee. Some of the catchwords used to describe the objectives of the reform were improved democracy and effectiveness, local solutions, integration, and service. Each district has a district manager but the middle management levels were organised differently. Some districts used a geographical subdistrict division, charging subdistrict man- agers with the task of finding local solutions to the integration of activities that were formerly managed by different functional departments, like between schools and kindergartens. Others maintained a functional structure on the district level and others again used target groups, like Children and Youth or the Elderly, to organise for integration.

The problem with the reform was that a very severe economic crisis struck the Swedish economy at the time the reform was launched. The City quickly found itself in financial difli- culties due to the fact that significant parts of the welfare system are managed by municipali- ties and unemployment reached levels that had been unheard of since the 1930s. The newly formed districts came under orders to cut their budgets drastically before they had a chance to organise properly. By 1993 the best managed districts had emerged from a very turbulent per- iod with their finances in order and an organiza- tional structure that might have to be pruned further because new cuts were to be expected.

The research team which has done the video taping of the meeting focused their interest on the role of the district managers in the reform process. Annual interviews had been conduc- ted with the 21 managers and measures of their conceptions of effectiveness etc., made. Large repeated survey studies of citizens, politicians, professional groups and personnel provided background. To be able to study the city man- ager at work, the research team was given per-

mission to videotape meetings of the top management group in a well managed district. In preparation for the interpretive work, all eleven members of the group had been subjec- ted to interviews on two topics; the (four-year) history of the district organisation and the group, and the roles of members in the group. Further background knowledge was based on the exten- sive studies of the districts over the previous years (Jonsson & Solli, 1994) and studies of the budgetary process of the city some 15 years earlier (Ionsson, 1982).

Since there were eleven members of the top management group we had 10 descriptions of each member. They were remarkably similar. For the purpose of this analysis we may note that the leader, A, was seen as dynamic and driving. There were also four female subdistrict managers described as talkative and some- times referred to as the four sisters. It should be noted that the participants have consider- able experience of management groups in the old, functional organisation. One of the advan- tages with the current group in comparison with the earlier functionally organised ones, was said to be that if you are a subdistrict manager and you bring up a problem there will be three other subdistrict managers who recognise the problem and can contribute. In the functionally organised groups of earlier experience everybody was a specialist and meetings (except budget meetings) would tend to be a sequence of dialogues between func- tional specialists and the boss. It was only nat- ural that the four subdistrict managers form an active subgroup. We called them the acti- vists. G, the opposition, is a committed social worker in charge of the Drugs and Prostitution Department. He speaks his mind and is likely to be the one to interfere when the group gets talkative and flimsy. A breath of fresh air! as somebody described him. Other persons in the group are the controller, B, a quiet and able person who usually sits to the left of the leader A. The other members, some of whom are generally less active and called the peripherals by us, are not active in the sequence that follows.

RELATE MANAGEMENT ACCOUNTING RESEARCH TO MANAGERIAL WORK! 425

The group functioned well together and nembers were proud of its achievements. It neets every fortnight to discuss and settle pol- cy matters and rules of the game. It is implicitly understood that these meetings are not for natters concerning individual departments. The whole district employs a little more than 2,000 people, with social services and schools dominating.

The method of data collection beside the preparatory interviews and the access to all documents was to videotape whole meetings. The videotape was then transcribed and the conversation was analysed in terms of utteran- ces, overlapping talk etc. From the video of the meeting seven sequences of about one to two minutes each were edited to a separate tape. This tape of sequences was then replayed, indi- vidually to each group member for comments. For each sequence, the respondent were asked to explain what goes on here? These post- meeting interviews we have called self-con- frontation interviews because part of the task for the respondent was to interpret her or his own performance. In all, we thus have 11 self- confrontation interpretations of each of the sequences. A:

The following sequence is the first on the edited tape. The first item on the agenda was Rents to charge when letting premises. The district has a lot of different rooms, from gyms and kitchens to conference rooms, that are not used for municipal activities all the time. What rent should be charged when letting those pre- mises? An officer had investigated and pro duced a memo that was presented to the group. The presenter had been interrupted by ques- tions and discussion several times. A crucial question that turned the discussion towards internal rents was Who gets the revenue? posed by one of the peripherals who did not participate in the following sequence. Some- body had introduced the example that a certain school (the Gustavi School) rented the gym from another school and would have to pay rent B: from their operating budget while the receiving @ school did not have the cost for the gym in its G: budget. Budget responsibility at this time did not c:

include rent, i.e. cost for premises was a central responsibility. Charging the Gustavi School rent for the gym would be sheer profit and no costs for the receiving school! On the other hand, responsibility should include all costs.

THE TEXT

Video-taped 1993-05-06, Sequence 13.06.30 - 13.09.10. (Summary on internal rents after a lengthy presentation of a memo on the pro- posed changes in the ground rules for charging rents.)

Personage: A = the leader B = the controller G = the opposition C ,D, I = activists. X = persons unidentified by the transcriber, [2 take it 12 = indication of overlapping talk (...> = indicates words that cannot be heard, usually because people are talking at the same time.

We will, [I we could] 1 start from how we basically treat premises, son of, and there the rule is that rents, base rents, are decoupled from budget responsibili- ty . ehhh...and if we then take...ehhh... this case with the school then, we do in fact pay full ehhh base rent, sort of independently of how it . ..ehhh... adds up in the Burred school or the Guldheden school. It is then reasonable that we pay full base rent, that is the rent for the gym of the Gustavi school as well, huh. You can argue that one should, sort of, start from that which is the . ..ehhh...base pre- mises for the activity. Then we could charge internal prices for the rest which are transactions. In that way...that would be a way to.... 11 yahll There is a silence for a long while. ehhh...I am not sure that I understand, but... no, me neither

S. JiiNSSON 426

I: D: A:

D: A:

B: I: A:

I: G:

A:

I: x: X: G: A: I: G:

I:

no [2 take it]2 once more P WI2 Well...ehhh...the basic rent for a premise is outside budget responsibility. That means that for the Guldheden school WE pay full rent for premises so to say also for the gymnastics activities. If we then come to the Gustavi school, one could say like this.. .ehh.. .they would pay for.. .ehh the gym and that would come out of the operating budget. m= = and that [3 would be]3 inside budget responsibility [3 ml3 =m= and then we would use a different budget principle for the Gustavi school than we have for the rest of the school activities... that . . . and that is not reasonable no no, thats not why (...) I agree about that but I am of the opinion Therefore I am of the opinion that [4 in this type of cases14 I believe that it is rea- sonable to argue [5 on this issue15 tha- t...ehh...we do not use internal charges, while we should do it in the rest of the cases, where this does not apply. [4 yes, it is]4 15 (...)I5 yes I [6 think that]6 [6 Shall we 16 [7 si]7 [7 (...)I7 [8 (...>I8 [8 (but what the hell)]8 I think that it can be a bit unfair (. . .) huh= = yes there are many other examples like this then where you find yourself [9puz- zledl9. I have been thinking about many. I could think of . . . in my office it would be the same effect. When we want to have a meeting at the office there is no room big enough to [ 10 take] 10 us all. Will I then have to pay every time I want to be somewhere for a meeting.. . or doesnt one have... to [ll others]11 so to say huh. What I mean is that; it will be [ 12

exactly] 12 the same stuff and there will be an [13 awful113 (...I

x: 19 (...>I x: [lo (...)I10 x: [ll mhmlll X: [12 mm 112 X: [13 mm 113 A: yes [14 we will take 114 I: [ 14 or is it ] 14 maybe, so hard to drive it

home, I dont know A: D! D: [15 ehhh 115 I: [ 15 yes ] 15 that is exactly what I think D: I also think we should have a special rate

for senior citizens.

WHAT IS GOING ON HERE?

The first reason we cut out this 2 minutes 40 seconds from the meeting of four hours was the long pause in the beginning where the picture seemed to freeze. It produced a comic effect. Then there was an outburst of demand for clar- ification with G, the Opposition, breaking the silence. Then there is a much clearer summary that focuses the problem illustrated by the Gustavi School paying from its operating bud- get for using the Guldheden school gym which was not responsible for the cost of premises in the first place. The proposed principle is that for planned regular use of premises there should be no internal rent charges, but for the rest, occasional use, as it were, which are called transactions, there should be a charge. G has something on his mind and tries to break in, but never gets to the point of articulation of his problem. Instead I, one of the activists, thinks of another example. When she has a meeting with her subdistrict management team there is no room big enough in the office and she usually holds the meetings in the neighbouring old age home where they have a nice room. That would mean the same kind of bureaucratic costs of charging rents. Then A, the leader, breaks out of the example giving exercise by giving the word to D, who comments on the original memo

RELATE MANAGEMENT ACCOUNTING RESEARCH TO MANAGERIAL WORK! 427

on external rents and proposes that there should be a special rate for senior citizens.

In terms of interaction in the group, A, the leader, is the dominating actor. He seeks to conclude a diversion on internal rents by ana- logy. He does not get it right and it is in accor- dance with Gs role that he breaks the silence of confusion. A gives it another try by relating the summary to the example in focus. Then I, an activist, thinks about other examples and the paper work it implies to have these internal charges. A marks irritation by saying yes (in English) and moves the meeting back on track with the help of D. Actor behaviour fits nicely into the roles group members have assumed over time. (We are not told what it is G is annoyed about, only that he thinks there is unfairness hidden somewhere).

We gain further insights into what was going on by examining the context of this exchange. There was a memo that had been circulated to participants the day before. In the meeting there was a presentation of the main points of that memo with questions, and there was the innocent question by H ; Who gets the rev- enue? This question alerts the participants to the added freedom of action (in a budget straight jacket) that rent revenue would give. Then comes the fairness question of the school with the gym (Guldheden) getting revenue from the Gustavi School without having responsibility for the costs of the gym.

This point on the agenda, which was the first one, ends with a decision to go to the council with a formal memo to request a decision on a price list for external rents. This is a rule of the game-only the council can decide on fees and charges to outsiders. Revenues would go to the activity centers, but if these grow exceptionally there will be a review of the rules. There are also a number of points in the meeting minutes about further development of the internal rents issue. The task is given to B, e.g. to devise administrative routines and the appropriate changes to the accounting system. The matter of rents is fragmented into 13 statements in the minutes. The original issue was external rents but the crux was the definition of territory and

revenue-sharing. The group discovers several new problems by solving the original one. It is not without consequences for internal relations to behave more business-like towards outsiders!

WHAT DID (S)HE MEAN? SELF-CONFRONTATION INTERVIEWS

The video tape containing this sequence (together with six other sequences from the meeting) was replayed to the participants in individual interviews three (summer-) months after the meeting. The structure of the inter- view was the following; each sequence was dealt with separately and the opening question concerning each sequence was What goes on here? The follow up questions on the first statement of interpretation varied depending on the specific circumstances that developed in the interview. In the following only those aspects of the interview that seem to add to the interpretation of the sequence are touched upon.

B, the controller (a peripheral), gives addi- tional perspective to the sequence. He says that the reason for this becoming a bit complicated is that the district managed to take only half the step when they designed economic responsi- bilities in the beginning of the reform period. There are rates for schools, child care, home service etc. (making resource allocation auto- matic and based on volumes). When they did this there was no time to work out rates for localities. They had to keep premises a cen- tral cost responsibility. Now there was an opportunity to take a further step towards more complete economic responsibility for the profitcenters/activity centers by including costs for premises. B did not see the sequence as complicated at first but, (R = researcher)

. R: But participants did not seem to under-

stand. A had to repeat before getting any reaction.

B: Well, 1 do not see that as indicating that people did not understand. We were in agreement on this issue.

428 S. JiiNSSON

R: Why did he have to repeat then? B: Well I suppose he wanted some kind of

response. The fact that I brings up (the example of paying for a meeting room) is for clarification.

B sees the incident as a natural continuation in the design of financial responsibilities.

The leader, A, also is of the opinion that there were no difficulties in understanding. This was not a complicated issue. When asked why he had to repeat then, he says, maybe members had not done their homework properly. A was better prepared. It is difficult to jump aboard if you are not prepared. The fact that I brought up the example with a meeting room is inter- preted by A as a questioning of the solution. The researcher says, when you say yes that means that you recognise the consequences of the rule and you realise that there may be pro- blems with it. A responds with maybe.

I , who brought up the case of the meeting room, admits that she did not understand the opening summary and she adds that it felt like nobody really grasped what he was talking about. When A repeated it was much clearer. It sometimes happens to all of us that we do not understand.

One cannot avoid the impression from the self confrontation interview that I has not really understood what the decision was. She under- stood the principles all right and she wants to increase her responsibility area to include inter- nal rents provided that there is money in the budget for it, but what about further decentra- lisation to lower levels?

G, the opposition, interprets the sequence as:

G: as usual A gives a long harangue. R: Dont you understand what he says? G: No not always, sometimes he speaks in a

way we dont understand. R: And then everybody is quiet? G: Yes it looks that way. R: And then several people ask for clarifica-

tion? G: Yeah, there is something there that isnt

right! Sometimes I think that it is some

bloody tactics of his just to get a decision through

R: He dominates a lot? G: Yes he does, but we let him!

In this sequence one might criticise the researcher for leading the respondent on improperly, but on the other hand we can judge that what G says is in line with his role in the group. It is not likely that the researcher would make him say things he did not mean to say.

Among other participants that did not take part in the exchange of the sequence we note that the record keeper of the group (a periph- eral) finds the matter easy to understand when the summary is repeated. The same goes for other participants. The example of the meeting room brought up by I was to confirm that she had understood and she got that confirmation. On the matter of not understanding what goes on the record keeper complains that her great- est trouble is that there often is no clear signal that a decision has been taken. It might be like this yes-statement in the sequence-and then you have to back up and try to interpret what the decision was. She usually tests her interpretations with the others and if A is avail- able she will check with him. Sometimes it is clear from the whole atmosphere of the meet- ing which way the group wants to go but it is still difhcult to formulate the decision. The group has discussed how to be clearer about decisions and there has been improvement. As record keeper and information officer, she can- not dose off like some of the others do some- times, but she likes being well informed and keeps alert through the meetings.

ANALYSIS

The purpose of this analysis will be to iden- tify two meaning generators uncovered by this approach to managerial work; the concrete case and the recipe. The first relates meaning to context, the second illustrates meaning related to role.

RELATE MANAGEMENT ACCOI;NTING RESEARCH TO MANAGERIAL WORK 42)

The concrete case The centre piece of this sequence is the role

of the concrete case of the Gustavi School paying internal rents for using the gym of the Guldhe- den School, should the proposed principle for internal rents be implemented inside the cur- rent responsibility system. On the basis of this case a first general principle is (not very cleari) it seems) formulated. A second version of the general formulation is clearer. This could be the end of the matter, because the original issue was external rents, but now a second candidate for concrete illustration is introduced by I. This is the case of regular but less frequent use of premises belonging to an activity centre and it is a higher organisational level (the sub district) using the premises of an activity centre. The matter of internal rents is much more compli- cated than asking the council to confirm rates for external rents! It has to be investigated fur- ther and it cannot be solved here at the table today since it involves budget allocations as well as principles for what is to be considered a transaction between responsibility centres. A cuts the discussion short and moves the meeting back to the agenda, already set on a decision to give somebody the task to prepare a memo on principles for internal rents. What is established in the sequence is a concrete case to be used as touch stone in formulating the rules for internal rents. The pronounced fairness criterion is that there should be a match between cost respon- sibility and budget allocation. The alternate candidate for test case, the meeting room, is cut short by yes, meaning that the rule would be generally applicable but this is not the time to come up with a lot of tricky cases.

One might interpret the work by the group as one of building a social representation of the system of rules for internal rents, with a con- crete case with a narrative appeal as the centre element. It should be noted that in order to understand the implications of the Gustavi School example one has to use cues in the meeting text outside the analysed sequence. A further contextual aspect was that the tit, had carried out a sale-lease back deal with a financial institution for a large part of the

municipal buildings (including most schools) the year before. This had lead to higher mar- ket based rents, and the cost of premises had become a problem which had to be dealt with. The district needed to find new revenue as well as reducing floor space.

The recipe A further approach to the sequence could

take as its point of departure the controllers statement that this was not a complicated issue, just a natural continued application of a successful recipe. The recipe to take the district through the difficult years of cost cutting was to decentralist economic responsibility based on clear rules of the game. Rates had been calculated for many activities establishing a transparent relation between volumes and resources. Managers of responsibility centres are able to calculate the consequences of their actions. The reason why the costs of premises were not included earlier was that the district did not have all of its floor space at its disposal all the time. The school byrns, for example. were let to the central Leisure department on a long term lease to allocate between sports clubs etc. Therefore a proper cost allocation would have to be preceded by a cost driver study. The rent contracted with the Leisure department could also be questioned on the basis of such a study, and priority for local clubs could be bargained for. However, the district did not have personnel for such a study and the matter of allocating responsibility for the cost of localities was left for later. Now the time had come to make the study and to include the cost for premises (which accounted for 17% of the districts budget) in the activity rates. Then cost centre managers would be induced to reduce floor space m,ithout extra pressure from top management. The recipe which had worked well in reducing costs for personnel and direct costs would work with premises as well. For the controller it was meaningful and natural to take this last step in perfecting the system of decentralised responsibility. It should be noted that this cue is discovered through the self-confrontation interviews and it is only

430 S. JiiNSSON

appropriate that B behaving in accordance with his role provided the clue for this meaning generator which is role dependent.

Generating meaning There seem to be at least two meaning gen-

erators at work in the group, one is the con- crete case of the Gustavi School which is used by members to test implications of proposed rules, the other is the recipe which is used by the controller and the leader to put the internal rent issue in frame. One makes sense in terms of a social representation, the other in terms of the vocabulary of an interpretive repertoire. We can expect the narrative to appear early in sense making processes and the vocabulary later. Accounting rules or responsibility speciti- cation can serve a bridging function, connecting the narrative with the vocabulary. A concrete case can be narrated and seen as application of the recipe. The recipe works by being able to accommodate the concrete case. A recipe has the instrumental characteristic that it links rules to ends. The concrete case has the narrative characteristic which makes it possible to manipulate the configuration of rules and see the implications, Both provide context to the internal rent issue. B, the controller, already has the vocabulary because of his role. He sees the concrete case as an opportunity to enlarge the area of application of the recipe.

After some further investigation of how the rules of the game should be designed and applied, the top management team will be able to integrate cost of premises into a joint social representation of managerial responsibility and accountability. As to the provision of added vocabulary to interpretive repertoires the recipe has the potential, being instrumental in character, to provide arguments in terms of (central austerity) goals and thus can serve in communication with the corporate level. The concrete case can serve to demonstrate the fairness of the new rules to activity centre heads. In both cases the communication will focus on giving meaning to management accounting rules in spe. The context will determine which meaning generator is appropriate.

DISCUSSION

This analysis of managerial conversation in a well functioning team illustrates how the same text offers different readings depending on the approach. The use of a concrete case to estab- lish meaning is illustrated by the use of the Gustavi School image as a touching stone (and the irritation over a competing case). This reading relates to the horizontal axis in Fig. 1 (Message - Context). The other reading sets the problem of internal rents in relation to the recipe of managing finances through decentra- lisation, subject to clear rules. This reading relates to the vertical axis in Fig. 1 (Sender - Receiver) and the role of the interlocutor B, who in his role as controller understands the situation in terms of further development of the management control system through the added specification of responsibility. Both readings make sense and are based in the recorded con- versation and the interpretations done in self- confrontation interviews (Participants seeing the sequence replayed and commenting on what goes on).

The epistemological status of these interpre- tations remains a problem. Clearly we have not analysed the conversation as such to extract recurrently used systems of terms . ..used in particular stylistic and grammatical construc- tions. organised around specific metaphors and figures of speech (Potter & Wetherell, 1987, p. 149) which is how interpretive reper- toires are described. Instead we have used the transcripts and interviews to detect cues (Potter & Wetherell, 1987, pp. 138ff) and then searched the context of the conversation and member roles for further building material. It might be ventured that this one case is too specific and narrow to allow any general con- clusion. Still management typically is conduc- ted in specific situations. In the Gilbert and Mulkay (1984) study, 34 scientists were inter- viewed and in that material, cues pointing towards the two repertoires at work appeared repeatedly. In this case further studies of meetings dealing with costs of premises and activity centre budgets can provide evidence of

RELATE MANAGEMENT ACCOUNTING RESEARCH TO MANAGERIAL WORK! 131

patterns in conversations and indicators of interpretive repertoires being used.

To ascertain evidence of social representa- tions seems more time dependent. The concrete case serves the team as an anchor for the for- mation of a touch stone in the deliberations on what rules are appropriate, but it is likely that the Gustavi School will be useful (be a case in point) for this purpose only temporarily. Evidence collected then and there will be rele- vant for analysis, but as the team moves on towards more elaborate articulations of the rules of the game other cases will be likely to serve the function of social representations even if the item on the agenda still is Internal rents.

The ontological status of the two meaning generators can be supported by references to relevant literature. The narrative character of the concrete case offers associative links in analogue descriptions of cases. The discussion in the meeting started to generate alternative cases (which the leader found premature and cut short). The concrete case allows the test- ing of how a principle would work (relate to the pragmatic truth concept (James, 1974; Baldwin, 1986)). Thinking through the conse- quences presupposes that the thinker has a thick description of the case. Controlling what case is used for this touchstone purpose will make it possible to control the outcome of such contemplation. Narratives have great power in normative discourse (Bruner, 1990).

The recipe is experiential (Kolb, 1984) in the sense that it accounts for a behavioural pattern, and instrumental in the sense that it summari- ses confirmed (successful) action. A recipe has a history and the specific associative link to consider here is the question of whether the definition of the current situation is similar enough (fits) to the experience condensed in the recipe. The recipe also has a teleological end. It informs us about what to do to achieve certain ends-what it is good for. Learning to apply the recipe includes experience of earlier applications as well as ability to determine a diagnosis of what immediate objectives have priority now.

SO WHAT WOULD IT MEAN TO ALIGN MANAGEMENT ACCOUNTING RESEARCH

WITH THE WORK OF MANAGERS?

The main consequence of orienting manage- ment accounting research towards the informa- tion managers use when they manage would be a focus on communication. This would alert us to the ontological aspects of the subject. Instead of assuming that management account- ing information goes directly into decision making models of some kind, we would have to assume that it comes to use in varying commu- nicative contexts. When Simon et al. (1954) talk about the attention directing function of accounting information they mean that accounting information may serve as an indica- tor that something is wrong. It is then necessary for the user to identitji relevant cause-effect relations, before a link between accounting and control can be established. Hedberg and Jiinsson (1978) advanced the idea that semi-confusing information could serve the function of alerting managers to the need to investigate what cause-effect mechanisms were at play. The case presented above is intended to illustrate the role of communication in such a focusing of attention. Simon et al. (1954) also mention problem solving as a context for the use of accounting information. In such a con- text we would expect to see efforts to connect recipes and concrete situations in social con- struction work. What are the chances t.hat par- ticipants from different levels will talk about the same thing in such situations? If not would we not need to develop a procedure for the analysis of many valued logic (i.e. implications of the type X is a manifestation of Y, Y is an attribution by B)? Are social representations temporary but interpretive repertoires more stable in the team and how is accounting data implicated?

The assumptions of rationality and opportu- nism, implicit in most decision making models, are sometimes criticised as not descriptive of real humans. If research attention is focused on managerial communication this may no longer be an issue. The question of rationality will be

432 S. J6NSSON

an empirical one. The observational data may support a rational model or it might be the case that some other model makes better sense of the observations. In both cases it is possible to conduct a constructive debate on the merits of theoretical statements against the background of empirical observation. This brings us back to the epistemological aspects of the kind of observation of managerial work exemplified in the case above.

THIS IS ETHNOMETHODOLOGY!

The observational method exemplified in the case above can be criticised in many ways, but it provides a solid empirical base. This conver- sation did in fact happen and researchers may go back and check again and again. It is not likely that the participants play-act to give an impression of their work which is not genuine. This is so because they are peers who are in a live situation where unprofessional behaviour will be corrected. The empirical study of how competent persons work to accomplish joint solutions which work in a complex situation falls under the label of ethnomethodology. What is needed is an elaboration of systematic analysis of data collected in the manner indica- ted above.

Ethnomethodology aims to understand folk (ethno) methods (methodology) for organising their world (Garlinkel, 1967). Those methods are laid down in the skills (artful practices) through which people develop, jointly, an understanding of each other and their social institutions. Bruner, (1990) also uses the folk approach in defining the narrative mode of cognition, and Buttny (1993) does the same for the study of accounts.

Ethnomethodology is not well defined and this has given rise to a number of misunder-

standings about its methods. Especially it is said that it is tarnished by voluntarism and subjec- tivity. This is a misunderstanding attributed to Bourdieu among others (Watson & Seiler, 1992, p. xiv). On the contrary ethnomethodologists agree on anchoring their analysis in texts of behaviour (i.e. not in assumptions about mental states). This has given rise to the label new empiricism. They also agree that social facts are not out there to be encountered, but are continuously constructed through interaction, which, contrary to mental states, is observable. Subjectivity is virtually eliminated by the fact that the registered text is a conversation between competent persons (Atkinson, 1988) in their area of competence and in a situation where their purpose is to reach a solution that works. The statements participants make are public and stand to be corrected by other competent par- ticipants. Their statements are empirical facts. What went on in their minds while they were making them cannot be registered. Researchers can come back to the empirical facts, the text, when discussing interpretations. Conversation and discourse thus allow and require interpre- tive feedback (note, in the case above, how A was obliged to repeat the proposed principle of accounting for internal rents) and thus provide for a negotiability of meanings which includes corrections and retrospective reinterpretations (Blimes, 1992, pp. 97). Participants reflect on what they are doing while doing it.

Ethnomethodologists are not interested in individuals or reading peoples minds, but in describing the methods people use when they generate accounts, toget