«Paul M. Collier Aston Business School, Aston University Accounting for Managers Accounting for Managers: Interpreting accounting information for ...»
The major contribution of cybernetics has been in the study of systems in which complexity is paramount (Ashby, 1956); it attempts to explain the behaviour of complex systems primarily in terms of relatively simple feedback mechanisms (Wisdom, 1956). There have been a number of attempts to apply cybernetic concepts to the issue of management control, but these have all been of a theoretical nature, albeit based on general empirical observation. The process of generating feedback information is fundamental to management accounting on which much management control practice rests, although this is not usually elaborated in any very insightful manner. However, Otley and Berry (1980) developed Tocher’s (1970) control model and applied it to organizational control. They maintain that effective control depends upon the existence of an adequate means of predicting the consequences of alternative control actions. In most organizations such predictive models reside in the minds of line managers, rather than in any more formal form, and they argue that improvements in control practice need to focus on improving such models. It is also argued that feedforward (anticipatory) controls are likely to be of more importance than feedback (reactive) controls.
Arguably the most insightful use of cybernetic ideas applied to management practice is also one of the earliest. Vickers (1965, 1967) applied many cybernetic ideas to management practice during the 1950s. Although developed in a primarily closed systems context, he also started to explore the issue of regulating institutions
RESEARCH IN MANAGEMENT CONTROL 279from a societal perspective. This is also a major theme of Stafford Beer, most comprehensively in his 1972 book Brain of the Firm. Here he uses the human nervous system as an analogy for the control mechanisms that need to be adopted at various levels in the control of an organization. However, Beer’s major contribution lies in his attempt to tackle issues of the overall societal and political context within which more detailed organizational forms and controls emerge. This is a theme which is picked up, albeit in a very different form, by the radical theorists of the 1980s and 1990s. The standard concepts of the cybernetic literature do not have such a straightforward application to the issue of organizational control as some presentations of them tend to imply. However, they do provide a language in which any of the central issues of management control may be expressed.
Further progress comes from the use of general systems theory which stresses the importance of emergent properties of systems, that is, properties which are characteristic of the level of complexity being studied and which do not have meaning at lower levels; such properties are possessed by the system but not by its parts. Systems thinking is thus primarily a tool for dealing with high levels of complexity, particularly with reference to systems which display adaptive and apparently goal-seeking behaviour (Lilienfeld, 1978). Some useful conceptual distinctions are drawn by Lowe and McInnes (1971) who attempt to apply a systems approach to the design of MCSs. An important extension to the realm of so-called ‘soft’ systems (i.e. systems which include human beings, where objectives are vague and ambiguous, decision-making processes ill-deﬁned, and where, at best, only qualitative measures of performance exist) has been made by the Checkland school at Lancaster (Checkland, 1981; Wilson, 1984).
One of the central issues with which the ‘soft’ systems methodology has to cope is the imputation of objectives to the system. In many ways this methodology reﬂects the verstehen (or insight) tradition of thought in sociology, where great stress is laid upon the accuracy and honesty of observation, the sensitivity and perception of the observer, and on the imaginative interpretations of observations. Although the soft systems approach has had considerable success in producing solutions to real problems, it does not appear to have contributed to the development of the theory of control in the normal academic sense. It is very much an applied problem-solving methodology in its present form rather than a research method designed to yield generalizable explanations, although it undoubtedly has further potential in this area. This raises the issue of the nature and type of theories that can be expected in such a complex area of human and social behaviour.
A framework to map developments in management control research Scott (1981) analysed the development of organization theory using two dimensions. First, he saw a transition from closed to open systems models of organization, reﬂecting the inﬂuence of systems ideas. Prior to 1960 most theorists tended to assume that organizations could be understood apart from their environments, and that most important processes and events were internal to the organization.
ACCOUNTING FOR MANAGERSAfter that date it was increasingly recognized that organizations were highly interdependent with their environments, and that boundaries are both permeable and variable. Second, he distinguished between rational and natural systems models.
The rational systems model assumes that organizations are purposefully designed for the pursuit of explicit objectives, whereas the natural systems model emphasizes the importance of unplanned and spontaneous processes, with organically emerging informal structures supplementing or subduing rationally designed frameworks. The distinction between rational and natural systems is applicable to both sides of the closed-open system divide, resulting in the deﬁnition of the four approaches.
We use these categories to summarize work in MCSs, although we recognize that such categorization is not necessarily ‘neat’. Organizations can be viewed as being both rational and natural, (Thompson, 1967; Boland and Pondy, 1983);
they are often intentionally designed to achieve speciﬁc purposes, yet also display emergent properties. It can be argued that each successive theoretical development provides an additional perspective which is helpful in understanding organizational processes, and which is likely to be additional and complementary to those which have preceded it. However, it is also recognized that this is a controversial statement that would not be accepted by some of those adopting a post-modernist viewpoint.
This work is characterized by being both The Closed Rational Perspective universal in orientation and systematic in approach, scientiﬁc management being a typical example. In the management control literature we ﬁnd a continuing emphasis on rational solutions, implicitly assuming a closed systems model of organization, which are universalistic in nature. Indeed, this can also be seen in much of the modern popular management literature, where a universal ‘how-todo-it’ approach continues to ﬁnd a ready market.
From a research point of view, there is much work which has sought to identify the ‘one best way’ to operate a control system. An excellent example of this approach applied to budgetary control is that conducted by Hofstede (1968). He sought to reconcile the US ﬁndings that budgets were extensively used in performance evaluation and control, but were associated with negative feelings on the part of many managers and dysfunctional consequences to the organization, with the European experience that budgets were seen positively but were little used.
Multiple perspectives are brought to bear, including systems theory, although this draws primarily on cybernetics. His conclusions list several pages of recommendations as to how budgets could be used effectively without engendering negative consequences, and indicate an implicit universalistic orientation. Similarly, the well known text co-authored by Anthony and a host of collaborators (see for example, Anthony et al., 1984) also clearly falls into the closed rational mode with its heavy emphasis on accounting controls.
in his article entitled ‘The Impact of Budgets on People’, an emphasis reversed neatly almost 20 years later by Schiff and Lewin (1970) in their article ‘The Impact of People on Budgets’. Lowe and Shaw (1968) discussed the tendency of managers to bias budgetary estimates that were subsequently used for control purposes.
Buckley and McKenna (1972) were able to publish a review article summarizing current knowledge on the connection between budgetary control and managerial behaviour in the early 1970s. Mintzberg followed his 1973 study of the nature of managerial activity with a 1975 study of the impediments to the use of management information, which dealt with many of the behavioural issues in the operation of control systems. There was thus a growing awareness of the human consequences of control systems use and operation beginning to emerge in the early 1970s, perhaps lagging some 20 years behind the equivalent human relations movement in the organization theory literature.
A behavioural perspective on the theme of managerial performance evaluation also began to emerge at this time. Hopwood (1972, 1974a) identiﬁed the different styles that managers could adopt in their use of accounting information and studied their impact on individual behaviour and (implicitly) organizational performance. Rhaman and McCosh (1976) sought to explain why different uses of accounting control information were observed, and concluded that both individual characteristics and organizational climate were signiﬁcant factors. A study by Otley (1978) yielded almost exactly contrary results to those of Hopwood (1972) because the research site had signiﬁcant differences; the conﬂicting ﬁndings could be reconciled only by adopting a contingent approach, a task that was more thoroughly undertaken by Hirst (1981).
The idea that systems used to evaluate performance are affected by the information supplied by those being evaluated has led to the concept of information inductance (Prakash and Rappaport, 1977; Dirsmith and Jablonsky, 1979). This generalized the observations of information bias and manipulation reported previously as just one manifestation of a more general phenomenon. Such work was extended by Birnberg et al. (1983) into a uniﬁed contingent framework, based on the ideas of Thompson (1967), Perrow (1970) and Ouchi (1979).
Despite the categorization of organizational contingency theorists into the open systems box by Scott (1981), the early contingent work in accounting-based control systems has a clear closed systems ﬂavour. It was only in the late 1970s that the open systems ideas in contingency theory, which followed primarily from the use of environment as a contingent variable, began to be reﬂected in the management control literature. This parallels developments in Organization Theory (OT), for it is arguable that early contingent work by writers such as Woodward (1958, 1965) concentrated on internal factors such as technology, and did not adopt an open systems approach until later. Thus, texts in the management control area, such as Emmanuel et al. (1985, 1990), Merchant (1985), and Johnson and Gill (1993) which recognized the behavioural aspects of MCSs as well as adopting some tenets of the contingency framework, tend to lie along the boundary of the closed natural category and the open rational approach.
As in OT, the emergence of an open systems The Open Rational Perspective perspective was accompanied by a return to more rational approaches and a
ACCOUNTING FOR MANAGERSrelative neglect of the natural (albeit) closed approaches of the preceding years.
The recognition of the external environment, key to the open systems approach, had never been strong in the early MCS literature. However, in the early 1970s there was a movement towards an open systems perspective, if only from a theoretical standpoint, an approach well illustrated by the collection of readings in the monograph New Perspectives in Management Control (Lowe and Machin, 1983). This approach was most cogently led by Lowe (1971) in an article which clearly recognizes the coalition of stakeholders involved in an enterprise (a concept used by Scott (1981) as an exemplar of an open systems approach) and the need for adaptation to the external environment. It is further clariﬁed in Lowe and McInnes (1971) article which adds the concept of resolution level. At the same time Beer (1972) was developing his own, somewhat idiosyncratic, approach and moving beyond cybernetics into more general systems analysis drawing heavily on neurophysiological ideas.
More signiﬁcant, empirically, was the development of the contingency theory of management accounting control systems (summarized by Otley, 1980). Although several contingent variables were shown to be signiﬁcant (e.g. technology, environment, organizational structure, size, corporate strategy), it was the impact of the external environment in general, and of external uncertainty in particular, that most clearly indicated the adoption of an open systems perspective. It is this distinction that marks the divergence of the study of management accounting systems, which have steadfastly retained their internal orientation (despite valiant attempts by a few proponents of so-called ‘strategic management accounting’ (Simmonds, 1981; Bromwich, 1990), and the study of the wider area of management control systems. Within management accounting, contingency theory waned in the early 1980s, to be replaced by various critical approaches in Europe, and to continue down a universalistic track in the US with the burgeoning popularity of activity-based costing under the leadership of Kaplan (1983) in particular. The wider study of control systems picked up on the neglected variable of corporate strategy at this point, which led to a small but signiﬁcant stream of work most notably by Govindarajan and Gupta (1985) and Simons (1987, 1990, 1991, 1995).