Firms, institutions and management control: the comparative analysis

Description
It is becoming increasingly recognized that management accounting and management control procedures and sys-
tems vary signi®cantly between organizations, sectors and societies. Four characteristics of control systems, in parti-
cular, di€er considerably between institutional contexts. These are: the extent to which control is exercised
overwhelmingly through formal rules and procedures, the degree of control exercised over how unit activities are car-
ried out, the in¯uence and involvement of unit members in exercising control, and the scope of the information used by
the control system in evaluating performance and deciding rewards and sanctions. These four characteristics can be
combined to constitute four distinct types of control system: bureaucratic, output, delegated and patriarchal. The
relative use of these kinds of control systemsÐand their e€ectivenessÐre¯ect major variations in the kinds of organi-
zations and ®rms that coordinate economic activities through administrative procedures, and their related institutional
contexts. The key features of ®rms here are the diversity of activities coordinated, their rate of change, shareholder
lock-in and the degree of owner management. These in turn re¯ect the nature of the ®nancial system and state struc-
tures and policies. Additionally, the ways that skill development is organised in a society and skills are controlled in
labour markets a€ect control techniques and practices, as do the nature of authority and trust relations.

Firms, institutions and management control: the comparative
analysis of coordination and control systems
R. Whitley
University of Manchester, Manchester Business School, Booth Street West, Manchester M15 6PB, UK
Abstract
It is becoming increasingly recognized that management accounting and management control procedures and sys-
tems vary signi®cantly between organizations, sectors and societies. Four characteristics of control systems, in parti-
cular, di?er considerably between institutional contexts. These are: the extent to which control is exercised
overwhelmingly through formal rules and procedures, the degree of control exercised over how unit activities are car-
ried out, the in¯uence and involvement of unit members in exercising control, and the scope of the information used by
the control system in evaluating performance and deciding rewards and sanctions. These four characteristics can be
combined to constitute four distinct types of control system: bureaucratic, output, delegated and patriarchal. The
relative use of these kinds of control systemsÐand their e?ectivenessÐre¯ect major variations in the kinds of organi-
zations and ®rms that coordinate economic activities through administrative procedures, and their related institutional
contexts. The key features of ®rms here are the diversity of activities coordinated, their rate of change, shareholder
lock-in and the degree of owner management. These in turn re¯ect the nature of the ®nancial system and state struc-
tures and policies. Additionally, the ways that skill development is organised in a society and skills are controlled in
labour markets a?ect control techniques and practices, as do the nature of authority and trust relations. Thus, Tay-
lorian control systems are unlikely to be widely used in countries where skill training is highly organised and controlled
jointly by employers and unionsÐas for example in many Central and Northern European states, just as delegated
ones are improbable in societies where systemic trust is low and authority patterns are patriarchal. # 1999 Elsevier
Science Ltd. All rights reserved.
Much of the traditional literature on manage-
ment accounting and management control was
founded on the belief that economic activities in
the developed industrial societies were organized
into clearly distinct and well-bounded corpora-
tions in which managers coordinated work and
sub-units through systematic rules and proce-
dures. The central task of the management control
system in these organizations was seen as ensuring
that work activities and sub-units ful®lled top
managers' objectives and provided the informa-
tion and systems to enable the managerial hier-
archy to correct any deviations from established
plans. In the words of Otley, Broadbent and Berry,
1995, p. 532), management control systems have
been understood by many to be ``a broad set of
control mechanisms designed to assist organiza-
tions to regulate themselves'' and include both the
provision of information about performance and
the implementation of corrective actions. In these
formulations the nature of the actors and objects
involved were typically taken for granted and
Accounting, Organizations and Society 24 (1999) 507±524
www.elsevier.com/locate/aos
0361-3682/99/$ - see front matter # 1999 Elsevier Science Ltd. All rights reserved.
PII: S0361-3682(97)00030-5
regarded as irrelevant to the design of the control
system. In particular, variations in: (a) who was
doing the controlling, (b) which set of interests
they were following, (c) what was the nature of the
organizations concerned and, (d) what was being
controlled, were usually ignored.
In most cases, the controllers were assumed to
be ``management'', i.e. the e lite group at the apex
of an administrative pyramid, who acted in the
interests of organizational eciency, or perhaps
e?ectiveness, as structured by the demands of
competitive capitalism. Organizations were taken
to be constituted by the legal and ®nancial norms
governing corporate forms in Anglo-Saxon capit-
alismÐor sub-units of theseÐwhich relied on for-
mal procedures to coordinate work activities
``within'' them. Such administrative integration
and control inside formal organizational bound-
aries was commonly, if implicitly, contrasted with
adversarial, ad hoc and anonymous coordination
of economic activities through market contracting
outside organizations, as is exempli®ed by the
contrast of ``hierarchies'' and ``markets'' in trans-
action cost economics. Finally, control was
usually exercised over task performance and the
organizational units responsible for conducting
economic activities within formal organizational
boundaries. Having structured the formal division
of labour following the usual contingencies of
size, technology and uncertainty, the managerial
control system was intended to monitor the
actual activities carried out against planned targets
and ensure the necessary corrective action could
be implemented. It was, then, typically viewed as
the necessary complement to the ``rational''
organization of work in a ®rm as decided by the
managerial team, and so followed the established
notions of rational management and organization.
In the past few decades, this approach has been
criticized for being too narrow, for assuming
managerial consensus over objectives, for taking
worker acquiescence and passivity for granted,
and for generating universalÐor at least highly
generalÐrecipes when it has become increasingly
clear that patterns of work organization di?er
greatly across sectors, regions and countries, as
does the nature of ®rms and economic actors
more generally, in ways that have signi®cant
consequences for how economic activities are
coordinated and controlled (e.g. Otley et al., 1995;
cf. Whitley, 1996, 1997). Not only has ``Fordism''
become transformed into ``Post-Fordism'', at least
in some authors' views (see, for example, Boyer,
1991, 1996; Shiomi and Wada, 1996), in ways that
a?ect the operation of control systems, but the
very di?erent ways of organizing work and eco-
nomic activities in general in Japan and other non-
Anglo-Saxon societies have become more widely
understood among Anglo-Saxon students of man-
agement control to the extent that a number of
studies relatng ``culture'' budgetary control proce-
dures and related phenomena have been published
in the past decade (e.g. Bailes & Assada, 1991;
Chow, Shields & Chan, 1991; Chow, Kato &
Shields, 1994; Chow, Kato & Merchant, 1996;
Merchant, Chow & Wu, 1995; O'Connor, 1995;
Ueno & Sekaran, 1991). Unfortunately, many of
these tend to posit a rather mechanical connection
between dominant norms and values in a society
and control practices which ignores the processes
by which the former impinge upon the latter.
In the light of the varieties of capitalism which
have become institutionalized in the late twentieth
century, and the consequent limited role of the
``Fordist'' mode of economic organization, it is
clear that models of management control premised
upon the ubiquity and superiority of this form of
capitalism need replacing with broader approa-
ches to deal with the considerable di?erences in
systems of economic coordination and control
across market economies. In particular, the var-
ious forms of ``¯exible specialization'', alliance
capitalism and business group networks highlight
the permeability of organizational boundaries and
the varied relations between ownership units,
administratively coordinated units and more
informally linked business groups in di?erent
societies (Hamilton, 1991; Langlois & Robertson,
1995; Westney, 1996). Coordination and control
of economic activities thus need not, and often do
not, occur only within formal bureaucratic hier-
archies under common ownership or by ``pure''
spot market transactions. Management control
involves extra-®rm linkages and commitments as
well as varying degrees of ``internal'' coordination
between organizational units in holding companies
508 R. Whitley / Accounting, Organizations and Society 24 (1999) 507±524
and diversi®ed Chinese family businesses (Hor-
owitz, 1980; Orru, Hamilton & Suzuki, 1989;
Redding, 1990).
Similarly, the extent to which and ways in which
owners, managers and di?erent groups of workers
are organized into distinct and separate interest
groups varies greatly between market economies,
and these di?erences a?ect both the division of
labour within organizations and the division of
organizational labour between them such that the
nature of controlling groups and the interests they
pursue are signi®cantly di?erent between societies
(cf. Kristensen, 1997; Whitley, 1997). Variations in
the structure and capabilities of unions and
employers' associations, for example, as well as in
the actions of the state, have had signi®cant con-
sequences for the adoption of ``scienti®c manage-
ment'' principles (Lazonick, 1990; Tolliday &
Zeitlin, 1991; Guillen, 1994). The study of man-
agement control therefore needs to take account
of variations in forms of economic organization
and interest groups, as well as in the societal
institutions and agencies that help to structure
them, in order to understand how and why di?er-
ent coordination and control systems become
established and widely di?used in di?erent insti-
tutional contexts.
As a preliminary contribution to such a com-
parative analysis of management control, this
paper outlines a framework for studying the con-
nections between key characteristics of control
systems, the nature of leading ®rms and organiza-
tions in a society and particular features of societal
institutions. These connections help to explain why
fully ¯edged Taylorism has only rarely become
established as the dominant mode of managerial
control in the 20th century, and under what con-
ditions we should expect ``responsible autonomy''
labour management strategies to be adopted,
amongst other phenomena. While not denying the
importance of technological and environmental
complexity and uncertainty in structuring control
strategies, this framework shows how broader
societal and organizational factors constrain and
guide the development and e?ective institutionali-
zation of distinct modes of management control.
First, I shall suggest a limited number of dimen-
sions for comparing characteristics of control
systems and how they can be combined to form
four distinct types. Second, I consider how these
characteristics are related to variations in work
organization and types of ®rms. Third, the key
in¯uences which a?ect the di?erential develop-
ment and use of these characteristics across
institutional contexts will be discussed, together
with their direct linkages to control system char-
acteristics. These concern the organization and
control of skilled labour power, ®nancial systems
and the state as well as fundamental norms
and conventions governing authority and trust
relations.
1. Characteristics and types of management
control systems
The general characteristics of management con-
trol systems which apply at both the work group
and organizational levels, and are interconnected
with particular kinds of work organization, ®rm
type and associated institutional contexts, can be
summarized for comparative purposes in terms of
four separate dimensions. First is their degree of
formalization, which is perhaps the most common
attribute of control systems which is discussed in
the literature. This is typically understood in terms
of quanti®ed, often ®nancial, indicators and mea-
sures and extensive reliance on codi®ed rules and
procedures. Often implicitly contrasted with direct
personal control over work activities and reliance
on personal, ad hoc, di?use and tacit evaluations
of performance, this dimension is, of course, a
standard component of bureaucratization indices
in comparative organizational analysis which
re¯ects the degree to which formal rules and pro-
cedures for coordinating activities and ensuring
compliance are generally relied upon in a society.
Here, a high degree of reliance on formal rules and
procedures implies a strong institutionalization of
impersonal regulations governing economic activ-
ities and their assessment.
At the work group level, this typically means a
preference for written instructions and guidelines
abut task de®nition and performance, including
the allocation of tasks, the selection of sta?, the
evaluation of outcomes and the ways that rewards
R. Whitley / Accounting, Organizations and Society 24 (1999) 507±524 509
are allocated. These reduce supervisor discretion
and ¯exibility in favour of general procedures and
relatively abstract rules. At the organizational
level, formalization refers to reliance on systematic
procedures for target setting, performance evalua-
tion and the implementation of corrective actions.
Personal and situational idiosyncrasies are dis-
regarded in favour of legalistic adherence to gen-
eral and relatively abstract rules in controlling
subunit activities.
Conversely, a low reliance on formal rules and
procedures implies greater personal discretion and
a tendency to take more features of the speci®c
situation into account in monitoring and evaluat-
ing performance. Here, formal rules may well exist
and be used, but idiosyncrasies of the people
involved and the particular situation are often
seen as being at least as important as the written
procedure. Standard setting, performance mon-
itoring and evaluation, and subsequent actions are
not dominated by formal procedures remote from
the speci®c circumstances and relationships involved.
As a key indicator of bureaucratization of
coordination and control systems, the degree of
formalization of control practices is often allied to
the second characteristic considered here, the
extent to which individual group and organiza-
tional unit behaviour is tightly prescribed and
controlled. This control can, of course, be exer-
cised either formally or informally, or both, and
the existence of formal rules and procedures about
task or unit performance need not imply highly
detailed monitoring of actual work processes.
While scienti®c management prescriptions implied
both a heavy dependence on formal procedures
for monitoring and controlling work with precise,
detailed instructions for carrying it out, many for-
mal measures of outputs do not specify how tasks
are to be conducted, or, in some areas of profes-
sional work, what the nature of the tasks is to be.
Equally, a preference for informal and personal
control over economic activities need not preclude
extensive supervision of work processes and
strong centralization of control over unit activ-
ities, as in patriarchal work systems (Redding,
1990; Silin, 1976; Whitley, 1992a, 1997).
A third dimension of management control sys-
tems which is often compared across organizations
from di?erent cultures is the extent of employee or
subunit involvement in target setting, monitoring
and evaluation of performance (e.g. Bailes &
Assada, 1991; O'Connor, 1995). While this would
normally be associated negatively with the extent
of tight control over subordinate behaviour in that
low discretion control systems are unlikely to
encourage much employee involvement, it does
help to distinguish two types of control system
where detailed prescriptions of task performance
are rather limited and open ended. Especially in
countries with strong skill based unions and an
e?ective, widely accepted training system, delega-
tion of control over the nature of work processes,
and how performance is to be monitored and
evaluated, to skilled workers may be quite con-
siderable in some industries (Kristensen, 1992,
1997). In contrast, output-based control systems
in, say Anglo-saxon and Japanese societies may
leave speci®c details of task performance to the
discretion of individuals and groups while in
practice deciding performance standards and indi-
cators quite centrally.
Finally, the scope and immediacy of feedback
and control mechanism vary considerably between
daily piecework based reward systems and more
long term career structures. While these typically
di?er between hierarchical levels in organizations
and between activities of di?ering complexity and
uncertainty, they also re¯ect more general features
of employment systems, labour markets and
®nancial systems. The stereotypical U.S. corpora-
tion, for example, tends to focus on relatively
short term and narrowly speci®c indicators of
performance in making career decisions and allo-
cating rewards whereas many large Japanese
companies evaluate blue and white collar sta?
over a number of years and in a number of roles.
They also tend to take more aspects of work per-
formance into account than simple, highly codi®ed
measures of results, especially overall contribution
and commitment (Aoki, 1988; Dore, 1973, pp.
110±113; Inohara, 1990; Rohlen, 1974). This
aspect can be termed the scope of the control sys-
tem and the information it relies upon.
These four characteristics of control systems
apply at both the individual/work group level and
the organizational subunit level. Other aspects,
510 R. Whitley / Accounting, Organizations and Society 24 (1999) 507±524
such as the relative importance of group or indi-
vidual performance and the extent to which the
interaction of subunits is systematically coordi-
nated and controlled, are often applicable more to
one of these levels than the other. In terms of
characterising distinctive kinds of control systems,
they can be combined into a wide variety of typesÐ
although not all combinations are empirically
likely as indicated above. However, four major
kinds which seem to occur quite frequently can
readily be distinguished. These can be termed the
bureaucratic, the output, the delegated, and the
patriarchal, which do, of course, have a number of
subtypes. Their key characteristics are summarized
in Table 1.
Bureaucratic control systems have been exten-
sively discussed in the organizational literature
and typically combine high levels of formalization
and reliance on written rules and procedures with
detailed instructions about how tasks and activ-
ities are to be performed, as in ``scienti®c man-
agement''. Such systematic and speci®c
prescriptions tend not to encourage subunit parti-
cipation in target setting and performance evalua-
tion so that this kind of control system is
essentially a top down one. Equally, it tends to
focus on highly speci®c information and perfor-
mance indicators derived from formal procedures
and so is relatively narrow in scope.
Output based control systems have been con-
trasted with bureaucratic ones in terms of their
much lower speci®cation of how activities are to
be carried out. While subunits may be more
involved in setting performance standards and
monitoring outputs than in bureaucratic control
systems, this is unlikely to be very signi®cant except
under conditions of strong collective organization
and high employer dependence on subordinate
commitment. Relying extensively on ®nancial
measures of performance to control activities
implies a lack of knowledge about how they are
performed and consequently rather limited ability
to assess technical competence and the potential
for improving work processes. Under these cir-
cumstances it seems unlikely that managers will
be willing to delegate substantial in¯uence over
standard setting and performance monitoring to
subordinate units. Finally, of course, this sort of
control system is narrow in its scope by de®nition.
Delegated control systems, in contrast, grant
considerable autonomy to subordinate groups and
units over work performances and may also
involve them in standard setting and monitoring.
Where organizations are fairly small and specia-
lized, control tends to be fairly informal and
direct, but more systematic procedures can be
expected to be established in larger ones, albeit
over longer time periods and more collectively
focused than in output control systems. Project-
based organizations in professional service indus-
tries, for example, often rely on fairly long term
and broad formal indicators of performance as
complements to high levels of skill development
and ``professional'' commitments to job perfor-
mance. In both ``artisanal'' and ``project'' or ``pro-
fessional'' types of delegated control systems,
evaluation relies on broad indicators of compe-
tence and contribution, often over years rather
than days, and presumes both high levels of tech-
nical expertise among employees and of their
commitment to maintaining it, as in Denmark
(Kristensen, 1992, 1997).
Patriarchal control systems are characterized by
much more personal and informal relations
Table 1
Control system types
Characteristics of control systems Bureaucratic Output Delegated Patriarchal
Extent of reliance on formal
rules and procedures
High High Mixed Low
Extent of control over how economic
activities are to be carried out
High Low Low High
Involvement and in¯uence of
subordinates in control system
Low Limited Medium Low
Scope of control system Limited Low High Considerable
R. Whitley / Accounting, Organizations and Society 24 (1999) 507±524 511
between controllers and controlled, with com-
paratively little reliance on written rules and pro-
cedures. This need not, of course, mean that these
do not exist but rather that managers, especially
the owner±manager at the top, prefer to rely on
direct supervision and personal contacts in mon-
itoring and controlling subunit activities, and
retain considerable personal discretion in making
decisions, bypassing established routines and
channels, as in many Chinese family businesses
(Redding, 1990; Silin, 1976). Typically, these sorts
of control systems grant little discretion to sub-
ordinates over task performance, nor do they
involve them much in standard setting or mon-
itoring. They tend to be highly centralized and
associated with considerable direction over work
processes, as in many Korean ®rms (Janelli, 1993).
This type of control system is, of course,
strongly connected to particular kinds of authority
relations, broadly summarized as paternalist, that
are widely institutionalized in many industrializing
societies. These subordination relations presume
an inequality in status and competence, and
sometimes morality, between supervisor and
supervised such that the former takes responsi-
bility for the interests of the latter (Beetham,
1991). Such control systems involve broad and
di?use indicators of performance and commit-
ment where personal connections are involved, but
it should be borne in mind that many employees in
such organizations are often relatively unskilled
and work for only a short time there, especially
female ones in labour intensive industries. Control
over work processes in such cases is usually highly
speci®c and short term (Deyo, 1989).
2. Control systems, work organization and ®rm
type
These di?erent kinds of management control
systems and their characteristics are clearly con-
nected to di?erent ways of organizing work and
managing employees, and to variations in the
kinds of enterprises that implement them.
Bureaucratic control systems, for example, tend to
be associated with quite rigid and specialized divi-
sions of tasks which are usually decided by senior
managers who have considerable discretion over
how work is to be organized. At least in the pure
scienti®c management model, supervisor±sub-
ordinate relations are typically remote and distant
since the latter are relatively unskilled and merely
required to carry out formal instructions. For simi-
lar reasons, organizations that establish these kinds
of control systems do not rely on the initiative and
commitment of their workforce and have few, if
any, commitments to their continued employment.
Labour management in these ®rms is typically
short term and heavily reliant on external labour
markets. Such control systems are usually devel-
oped in large companies which are often vertically
integrated and not highly diversi®ed across indus-
trial sectors since this would limit the ability of top
managers to control how economic activities were
to be carried out.
As this example indicates, the adoption of par-
ticular control systems re¯ects general managerial
choices about work organization and labour
management policies, as well as the nature of the
managerial elite itself, its relations to major own-
ers and lenders, and its dependence on the success
of individual enterprises and industries as distinct
from its identi®cation with managerial expertise
and ``managers'' in general (cf. Stewart, Barsout,
Kieser, Ganter & Wulgenbach, 1994). Management
control systems are particularly closely connected
to the prevalent division of labour, task allocation
and coordination and the sorts of employment
strategies followed by ®rms. Three aspects of these
broad features of work organization are particularly
important for control system characteristics. First,
the fragmentation and rigidity of task de®nitions
and allocation. Second, the extent to which coor-
dination of work is exclusively hierarchical or
relies on some horizontal integration between
work stations by employees. Third, the willingness
of managers to rely on the skills and commitment
of their core manual and white-collar workforce in
developing competitive organizational capabilities,
and their corresponding commitment to retaining
them over business cycle ¯uctuations.
Turning now to consider more general features
of ®rms and organizations, in addition to the
overall size of a ®rm being closely related to the
tendency to formalize control proceduresÐwhich
512 R. Whitley / Accounting, Organizations and Society 24 (1999) 507±524
itself depends on broader institutional norms
dealing with trust and authority as the Korean
chaebol demonstrate (Janelli, 1993; Kim, 1992;
Steers, Shin & Ungson, 1989)Ðthere are at least
four further ones connected to control system
characteristics. These concern the variety and rate
of change of their basic activities and capabilities
and the nature of owner±manager relations,
commitment and risk sharing.
First, the diversity of economic activities coor-
dinated by ownership units di?ers signi®cantly
between business systems and a?ects the choice of
feasible control systems. Second, the extent to
which ®rms alter their basic capabilities and
activities by, for example, readily buying or selling
subsidiaries varies greatly between, say, Japan,
Taiwan and the U.S.A. (Fligstein, 1990; Hollings-
worth, 1991; Whitley, 1992a). Third, shareholders
and bankers vary considerably in the degree to
which theyÐand top managersÐare e?ectively
locked-in to the destinies of particular enterprises
such that they cannot easily ``exit'' from custo-
mers. Such lock-in increases risk bearing and
sharing between investors and managers, and
clearly encourages closer coordination and control
of ®rms' activities. Finally, the extent of direct
owner control of managerial decisions and invol-
vement in day to day activities usually has quite
strong connections with how they control them,
especially in institutional contexts which trust in
formal procedures is low and authority relations
are more paternalist rather than contractarian.
Typically, the combination of low trust cultures
and paternalist political systems encourages direct
owner control and the use of patriarchal control
systems (Fukuyama, 1995). Conversely, delegated
control systems are often associated with much
broader job speci®cations and more ¯exible use of
skills, closer ties between supervisors and sub-
ordinates based on common expertise and training
and greater employer reliance on maintaining a
highly skilled workforce and commitment to high
task performance. They tend to be used by ®rms
that are not highly diversi®ed or liable to radically
change their basic technologies or markets and
that are able to share some of the risks of remain-
ing in single or related sectors with shareholders
and banks.
These sorts of general connections can be fur-
ther di?erentiated by considering the four char-
acteristics of control systems outlined above in
relation to a number of key characteristics of
organizations before analyzing how these are all
related to speci®c features of their institutional
environments. Considering ®rst the connections
between control system characteristics and pat-
terns of work organization, two major ones deal
with the division of labour and coordination
mechanisms. As summarized in Table 2, a frag-
mented and rigid division of tasks and roles is
typically associated with a high reliance on formal
rules and procedures and extensive prescription of
Table 2
Connections between control system characteristics, work organization and types of ®rm
Control system characteristics
Reliance on
formal rules
Close control
over behaviour
In¯uence of
subordinates
Scope
Work organization
Fragmentation of tasks + + À À
Hierarchical coordination of activities + À
Dependence on employee skills and commitment À + +
Firm type
Size +
Diversity of activities coordinated À À À
Discontinuous change in activities À À À
Major shareholder/bank lock-in +
Owner control À + À +
R. Whitley / Accounting, Organizations and Society 24 (1999) 507±524 513
how tasks are to be done because tasks are
simple and routine and can be largely speci®ed by
written instructions. Clearly subordinates are
unlikely to be involved in the operation of the
control system in this situation and performance
is fairly narrowly monitored and evaluated. Hier-
archical coordination of work activities may be
carried out either by formal rules or through more
personal means, but typically it implies close
supervision of work processes in order to ensure
they are e?ectively integrated where horizontal
connections are not encouraged. Again, sub-
ordinate involvement in the setting of standards is
not common here, but the scope of the control
system could be either narrow or broad depending
on authority relationships.
Focusing next on the connection between con-
trol systems and labour management strategies,
it seems likely that the more managers depend on
employees' skills and long term contribution, the
less likely they are to prescribe how work should
be done and the more they will involve sub-
ordinates in the operation of the control system,
especially where workers are strongly organized.
Similarly, high levels of employer±employee inter-
dependence will encourage managers to be more
concemed with broad assessments of individuals'
and groups' contributions than narrow perfor-
mance indicators.
Turning to the relations between control system
characteristics and overall ®rm and organizational
features, the most commonly noted one is that
between size and formalization. This is, though,
a?ected by skill levels and uncertainty, as well as
broader cultural conventions. Size, per se, does
not seem to be directly linked to the other three
characteristics considered here. Diversity of activ-
ities and capabilities, however, has wider implica-
tions. Where ®rms are engaged in quite diverse
industries, they will ®nd it dicult to prescribe
how di?erent subunits are supposed to carry out
their activities given their relative lack of knowl-
edge of widely di?ering technologies and markets.
Reliance on general, often ®nancial, indicators is
therefore more likely here than when diversi®ca-
tion is limited. However, because information
from di?erent subsidiaries is limited in its detail, it
seems improbable that top managers will delegate
much in¯uence over target setting and perfor-
mance monitoring to subunitsÐalthough, of
course, they may not do so either in less diversi®ed
®rms. The operation of the control system, then, is
likely to remain quite centralized in multi-sector
enterprises. For similar reasons, the scope of con-
trol systems is quite narrow in these kinds of ®rms
given the lack of detailed information that can be
standardized in evaluating speci®c performance
contributions across widely di?ering technologies
and markets. Anglo-saxon conglomerates thus
typically rely heavily on highly formalized ®nan-
cial control systems (Horowitz, 1980).
Third, ®rms which undergo radical dis-
continuities in the nature of the activities they
engage in and of their organizational capabilities
are also more likely to rely on general ®nancial
indicators of performance than engage in systemic
monitoring of detailed behaviours and decisions.
Similarly, they are unlikely to delegate much
in¯uence to subsidiary managers over how the
control system operates given the considerable de
facto decentralization of operational decisions and
the lack of attachment between the unit of owner-
ship coordination and the separate operating
organizations. Where ownership is readily trade-
able so that the organizational competences of
®rms can change quite radically as businesses are
bought and sold in an active market for corporate
control, commitment between top owners±man-
agers and subsidiary managers will not be high
and so controls are more likely to be imposed
from the top than be negotiated between
managerial colleagues. The lack of stability in
technologies and markets will also reduce the
likelihood of using long term, broad indicators of
performance in favour of narrowly focused con-
trol systems, as many large Anglo-Saxon diversi-
®ed companies demonstrate.
On the other hand, where, fourth, major share-
holders and creditors are locked into the fate of
individual enterprises and cannot easily ``exit''
from their commitments, but are not directly
involved in management, they are likely to insist
on both formal procedures governing major deci-
sions and frequent ¯ows of both formal and
informal information to manage the greater risks
involved. In general, the discretion of top man-
514 R. Whitley / Accounting, Organizations and Society 24 (1999) 507±524
agers over strategic choices will be less than in
situations where owners are easily able to sell their
shares on a liquid secondary capital market
because the interdependence between ®nancing
institutions and ®rms is greater and is more likely
to be controlled through systematic written rules.
This, in turn, will encourage a greater reliance on
formal procedures within enterprises as managers
seek to ensure their control in a corresponding
manner. It need not mean, though, that such rules
prescribe how operations are to be managed, nor
does it rule out mutual determination of how the
control system will operate, given the relatively
long term and interdependent nature of the rela-
tionship.
It will, however, encourage a more elaborate
and wide-ranging control system being developed
in most ®rms as planning and control of economic
activities can be more long term and assume a
greater degree of risk sharing between investors
and managers. This interdependence does, of
course, itself lead to greater concern with detailed
target setting and evaluation of opportunities and
outcomes over longer periods as managers and
bankers are committed both to each other and to
particular industries. Because they cannot readily
exit from a particular ®rm or industry, they will
want to be well informed about ®rms' activities
and performance to manage the greater risks
involved than if they could simply sell their shares
or move elsewhere. The prevalent pattern of alli-
ance capitalism in Germany and Japan manifest
these sorts of connections (Gerlach, 1992; Lane,
1992).
Where owners directly control enterprises, in
contrast, they will prefer to retain considerable
discretion over decisions and are unlikely to wish
to be subjected to the constrains of formal rules
and procedures. They are more likely, though, to
exercise strong control over subunit performance
and to supervise directly subordinate activities,
especially in societies where the level of trust in
formal institutions is low and subordinate com-
mitment to organizational goals dicult to ensure,
as in many Chinese family businesses (Silin, 1976;
Redding, 1990). Although a?ected by skill levels
and the collective organization of employees,
owner managers will retain considerable in¯uence
over how control systems operate and are also
likely to rely on a wide variety of information in
seeking to control subunits. Given their general
unwillingness to delegate control and rely on
indirect sources of information, control systems in
such ®rms are unlikely to be narrowly focused.
3. Control systems, ®rms and societal institutions
These connections, and the nature of ®rms and
markets more generally, are also linked to broader
features of the socio±economic context in which
economic activities are embedded. In particular,
the cultural conventions governing trust and
authority relationships in a society, the nature of
the ®nancial system and state±society relations,
and the ways that skills are developed, certi®ed
and controlled all a?ect the division of labour
within organizations, the division of economic
coordination between organizations and ®rms and
the ways that ownership is organized and related
to management (Whitley, 1992b, 1997). Some of
these phenomena also directly in¯uence control
system variables, such as the dependence of formal
control procedures on more general adherence to
formal rules and procedures governing subordina-
tion relations and contractual compliance. In com-
paring variations in control systems, then, these
societal features are of critical importance.
At least six major features of organizations'
institutional contexts are likely to a?ect directly
the sorts of control systems they develop. These
are listed in Table 3 with their links to particular
characteristics of control systems. First of all, two
sorts of broad cultural conventions governing
social relationships are important for the coordi-
nation and control of economic activities: trust
and authority norms. As has been emphasized in a
number of recent discussions, the ways in which,
and extent to which, societies generate trust
between business partners in economic transac-
tions are crucial to how industrial capitalism
develops and the forms it takes (e.g. Zucker,
1986; North, 1990; Sako, 1992; Fukuyama, 1995;
Lane & Bachmann, 1996). In particular, the
extent to which economic actors feel able to rely
on formal procedures and institutions in making
R. Whitley / Accounting, Organizations and Society 24 (1999) 507±524 515
commitments to business partners and ensuring
their competenceÐas distinct from relying on direct
personal knowledge or recommendationsÐa?ects
their willingness to delegate control to inter-
mediaries and rely on procedural measures to con-
trol their behaviour. This, in turn, of course
in¯uences their ability to establish large organiza-
tions controlling complex technologies and devel-
oping distinctive organizational capabilities.
Relatedly, authority relationships vary greatly
in their reliance on formal rules and procedures
and their recognition of subordinates as formally
equal parties to employment agreements. Pre-
dominantly paternalist patterns of subordination,
for instance, treat employees as being in the care
of superiors who are supposed to act in their best
interests given the limited capacity of subordinates
to do so (Beetham, 1991, pp. 88±90). In contrast,
more contractarian norms of authority treat both
superior and subordinate as being, in principle,
equal citizens and freely able to enter into con-
tracts on the basis of their own equally valid per-
ception of their best interests.
Where both trust in formal institutions and
procedures is low and authority relations are more
paternalist than contractarian, as in many non-
Western societies, control systems are likely to rely
more on personal supervision than on formal
procedures and to grant superiors considerable
discretion in their operation. Subordinates will be
tightly monitored and controlled since, in general,
trust in their competence and commitment is unli-
kely to be high, except for those who have won the
personal commitment of the owner manager.
Similarly, subordinates will have little in¯uence on
the way the control system operates and the eva-
luation of their performance will be wide-ranging
and involve a considerable variety of indicators,
including personal ones.
While state structures and policies in general
have considerable in¯uence over the sorts of ®rms
and markets that develop in a particular society,
and the ways they organize work systems (Whit-
ley, 1996, 1997), there are two particularly impor-
tant features which directly impinge upon the
control system characteristics discussed here.
These are: (a) the extent of business dependence
on the state and, (b) the degree to which economic
activities and coordination are formally organized
through the legal system and state regulations.
Business dependence on the state re¯ects the
extent of state dominance of civil society, on the
one hand, and its involvement in managing eco-
nomic activities and sharing investment risks, on
Table 3
Direct connections between societal institutions and control system characteristics
Control system characteristics
Institutional features
Reliance on
formal rules
Close control
over behaviour
In¯uence of
subordinates
Scope
Cultural conventions
Low Trust in formal institutions
and procedures
À + À +
Paternalist authority relations À + À +
State structures and policies
Business dependence on the state + À
Extent of formal regulation of markets
and economic coordination
+
Skill development and control
Strength of public training system and
of employer-union collaboration
À +
Strength of unions and of horizontal
interest groups in general
À +
516 R. Whitley / Accounting, Organizations and Society 24 (1999) 507±524
the other hand. In both South Korea and Taiwan,
for instance, the state has been dominant since the
end of the Second World War, with little if any
alternative sources of collective mobilization and
in¯uence, but the Taiwanese state has largely
remained aloof from the privately owned export
oriented sector while the Korean state has been
extensively involved in directing investment and
controlling industry entry and exit (Jones &
Sakong, 1980; Gold, 1986; Wade, 1990; Woo,
1991). For most Taiwanese SMEs in the export
tradeÐand most exporters in Taiwan remain
relatively small in sizeÐtheir dependence on the
state is, then, lower than it is for their counter-
partsÐwho are much larger as a result of state
policiesÐin Korea (Levy, 1991; Fields, 1995).
Where business dependence on the state is as
high as it is in Korea, such that political risk is as
great, if not greater than market risk, ®rms are
likely to be highly centralized and to exert con-
siderable control over the behaviour of subunits.
Correlatively, subordinate in¯uence on target set-
ting and performance monitoring is quite weak,
not least because these are often determined by
state agencies in highly state dependent businesses.
Overall, then, strong state involvement in the
economy and in ®rms' strategic choices is likely to
encourage the use of top-down, prescriptive con-
trol systems where countervailing in¯uences are
weak and poorly organized.
States also vary greatly in the extent to which
they directly or indirectly regulate markets and
industry structure through formal rules and pro-
cedures, as well as through various corporatist
mechanisms. Anglo-Saxon capitalisms by and
large engage in little such formal regulation, while
many central European and Scandinavian states
have elaborate licensing rules and encourage trade
associations and similar collective bodies to orga-
nize industry development. The more such formal
regulationÐwhich need not imply strong state
directionÐof economic activities exists in a
society, the more we could expect ®rms to institu-
tionalize formal control systems in their own
organizations as they adapt to and imitate key
features of their environments.
Turning to consider the skill development and
control system, the ways that practical skills are
acquired, certi®ed and organized have been shown,
principally by the Aix group and their colleagues
(Maurice, Sorge & Warner, 1980; Maurice, Sellier
& Silvestre, 1986), to have important connections
to di?erent patterns of work organization and
authority structures in Europe. Relatedly, the
ways that unions are organized, and the extent of
their in¯uence, a?ect managers' ability to struc-
ture the division of labour and authority in orga-
nizations as well as the way in which they control
work processes (Kristensen, 1992, 1994, 1997;
Whitley, 1997). While some U.S. ®rms sought to
destroy craft workers' control over production in
the engineering industryÐsuccessfully according
to Herrigel (1993, 1994)Ðallied to the develop-
ment of mass production strategies, in some Eur-
opean countries the unions were able to prevent
such ``rationalization'' (Sorge & Maurice, 1993)
and, in Denmark at least, have maintained sub-
stantial craft control over skill development and
allocation (Kristensen, 1997).
The speci®c features of the labour system in
various countries which directly impinge upon
control systems can be summarized as two
dimensions: (a) the overall strength of the public
training system and the extent of union±employer
collaboration in its operation, and, (b) the collec-
tive strength of the unions and of horizontal
interest groupsÐsuch as professional associa-
tionsÐin general. The public training system can
be considered to be strong when it develops stan-
dardized skills that are well regarded by both
employees and employers, and are crucial for
entering a wide range of labour markets, including
those for non-manual work, as in, say, Germany
(Lane, 1989). Union strength refers to member-
ship density, organizational cohesion and political
in¯uence as well as the ability to a?ect workplace
relations. Horizontal interest groupsÐespecially
those based on certi®ed expertiseÐare important
in¯uences on organizations because they a?ect the
discretion of ®rms to structure the division of
labour and control work de®nition and evaluation
(Child, Fores, Glover & Lawrence, 1983; Kris-
tensen, 1994; Botti, 1995).
Where the public training system is both strong
and involves employers and unions collaborating
in its management with the state, it produces skills
R. Whitley / Accounting, Organizations and Society 24 (1999) 507±524 517
that are relatively highly valued by traineesÐand
so attracts high quality applicantsÐand by
employersÐand so encourages ®rms to trust the
competence of skilled workers. It also provides the
basis for union±employer cooperation in other
spheres. In such societies it seems more likely that
managers will be less concerned to exercise tight
control over employee behaviour and will also be
more willing to allow some employee role in the
operation of the control system, especially if they
are themselves dependent on the fortunes of par-
ticular ®rms and industries (Whitley, 1997), than
where the training system is weak and union±
employer relations adversarial.
Relatedly, strong unions, and horizontal interest
groups in general, will limit the ability of managers
to impose tight controls over worker behaviour
and also encourage them to share some in¯uence
over standard setting and performance evaluation.
This is particularly probable when skills are highly
regarded by employers and there are limits on
their ability to deskill work processes. Thus,
``bureaucratic'' control systems are less likely to be
widely introduced, or to prove e?ective, in socie-
ties where both the training system and the unions
are strong, relative to more ``delegated'' ones.
These connections between control system
characteristics and dominant societal institutions
can be combined with those between ®rm char-
acteristics and control systems to develop some
tentative suggestions about the conditions in
which the four types of control system identi®ed
earlier are likely, or not, to become widespread in
particular sorts of ®rms in di?erent societal con-
texts. Since ®rms vary systematically in their
diversity, ¯exibility and ownership relations
between market economies because of these insti-
tutional variations (Whitley & Kristensen, 1996),
the actual number of possible combinations of
these characteristics likely to become institutiona-
lized in capitalist societies is more limited than
might at ®rst be thought.
These interconnections mean that control sys-
tems which are more likely to be e?ective in a
particular kind of ®rm will often also tend to be
used in certain institutional contexts which
encourage those sorts of ®rms and their associated
control systems to develop. The impact of ®rm
type and institutional features on the implementa-
tion of particular control systems is thus com-
bined. This is especially signi®cant when these
features discourage the adoption of one of these
types. As Table 4 suggests, for instance, low trust
in formal institutions and procedures inhibits reli-
ance on bureaucratic control systems because of
the perceived unreliability of written rules. It also,
though, encourages strong owner control and thus
additionally makes it unlikely that full scale
bureaucratic systems will be relied upon in such
societies.
Table 4
Connections between control system type, ®rm characteristics and societal institutions
Control system type
Bureaucratic Output Delegated Patriarchal
Firm characteristics
High diversity of activities À + À
Discontinuous change À + À
Shareholder/bank lock-in +
Owner control À À À +
Size À
Societal institutions
Low trust in formal institutions À À À +
Paternalist authority relations À À À +
High business dependence on the state À À +
Strong formal regulation + À
Strong public training system À + À
Strong unions À + À
518 R. Whitley / Accounting, Organizations and Society 24 (1999) 507±524
As Table 4 indicates, it is easier to suggest
negative connections between particular char-
acteristics of ®rms and institutions in a society and
the likelihood of managers adopting di?erent
kinds of control system than to be clear about the
conditions in which, say, bureaucratic ones will be
widespread. This is because the number of factors
which can inhibit or limit the use of any one of the
four characteristics which constitutes a control
system is considerable, and the institutionalization
of one of these four idealized systems usually
requires a combination of features which may
occur infrequently. It should also be noted that
many of these connections are based on relatively
extreme values of characteristics and often they
are not reversible. While, for example, diversi®ed
conglomerates rarely impose fully bureaucratic
control systems on their subsidiaries because of
the diculty of monitoring and prescribing beha-
viours in quite di?erent industries, it does not fol-
low that more focused enterprises tend to
implement such systems since this would depend
on other conditions, such as trust and authority
patterns and the nature of the labour system.
Considering next the conditions in which these
four types of control systems are likely to become
widely established in a society, the widespread
implementation of highly bureaucratic control
systems combining formalized rules and proce-
dures with detailed control over work behaviour,
task fragmentation, etc. as in scienti®c management
prescriptions is limited to a small number of ®rms
and societies, as indeed the wholehearted adoption
of ``Taylorism'' and ``Fordism'' has proved (Boyer,
1991, 1994; Guillen, 1994; Shiomi & Wada, 1996).
It is only feasible as a reliable means of controlling
work and employees in societies which combine:
(a) a widespread acceptance of formal rules and
procedures for regulating economic behaviour and
subordination relations, (b) markets which are
mostly organized around impersonal and formal
regulations, (c) weak unions, (d) an ine?ective
public training system, (e) low levels of owner
control, and hence considerable managerial
autonomy, and, ®nally, limited diversi®cation and
change of activities amongst leading ®rms.
Vertically integrated mass production enter-
prises which dominate their markets in legalistic,
capital market based, business dominated societies
are, perhaps, the only ones where such control
systems could be e?ectively adopted, as in some
sectors of U.S. industry in parts of the twentieth
century (Lazonick 1990; Campbell, Hollingsworth
& Lindbergh, 1991). Elsewhere, formal controls
are supplemented by more personal ties and
superiors are not so constrained by rules and pro-
cedures and/or the extent of control over the work
process is much less, with more discretion being
granted to subordinates.
Output control systems also require an ability to
rely on formal rules and procedures to control
task performance, but much less prescription over
how they are carried out. They therefore are used
in societies with e?ective legal and contracting
systems where authority is more contractual than
paternalist, and with varying types of labour sys-
temÐunless a strong union movement limits radi-
cal changes in organizational activities and
capabilities. Diversi®ed, rapidly changing ®rms are
more likely to rely on them, unless they are owner
managed. Thus, many holding companies that
move in and out of industries fairly quicklyÐ
which in turn usually implies a liquid market for
corporate control for buying and selling busi-
nessesÐrely on such control systems, especially in
capital market based ®nancial systems with well
established ®nancial and legal intermediaries.
Delegated control systems, on the other hand,
are encouraged by a strong public training system
and strong unions, as well as shareholder/bank
risk sharing as in many credit based ®nancial sys-
tems (Zysman, 1983; Cox, 1986). Essentially, they
involve managers sharing risks with core employ-
ees and so are more likely to be adopted when risk
sharing between di?erent groups and organiza-
tions is more institutionalized through mutual
lock-ins and strong social conventions encoura-
ging cooperation rather than adversarial competi-
tion. The more managers and skilled workers
come to believe they share a common dependence
on the success of particular enterprises and indus-
tries, the easier it will be to implement delegated
control systems. Corporatist societies with limited
managerial mobility, some bank risk sharing and
highly e?ective training systems are therefore more
likely to institutionalize these kinds of control
R. Whitley / Accounting, Organizations and Society 24 (1999) 507±524 519
systems, than those with dominant capital markets,
``ecient'' external labour markets and remote,
adversarial relations between state agencies,
employers and unions, especially where ®rms are
not highly diversi®ed. Scandinavian and Central
European countries seem to have developed these
kinds of control systems more widely than Anglo-
Saxon ones or those, such as France, where the
state retains considerable control over leading
sectors of the economy, unions are weak and
owner control is extensive (Lane, 1989; Kristensen,
1997; Lilja, 1997).
Finally, patriarchal control systems are pre-
valent in societies where trust in formal rules and
procedures is low, authority is typically paternal-
ist and business dependence on the stateÐand the
overall centralization of the economyÐis high.
Relatedly, owner control over ®rms tends to be
high and both the public training system and
unions are relatively weak. Because of the weak-
ness and perceived unreliability of formal
institutions personal trust is more signi®cant than
``systemic trust'' (Luhmann, 1988) and this limits
the size of complex organizations integrating var-
ied skills and capabilities since delegation of con-
trol through formal procedures is restricted. When
formal controls do exist in such situations, they
are typically combined with, if not dominated by,
informal obligations and commitments, and rarely
constrain the behaviour of superiors, especially the
owning family.
4. Conclusions and implications
This brief discussion of some connections
between management control systems types of
®rms and institutional contexts highlights their
variability across ®rms, sectors and societies, and
suggests some reasons for these di?erences.
Essentially, control system characteristics are seen
here as being closely linked to broader pattems of
work organisation and labour management stra-
tegies, as well as to variations in the nature of
®rms and the logic underlying their strategic
decisions. These phenomena in turn re¯ect the
societal contexts in which economic activities are
carried out, some features of which also directly
a?ect the e?ectiveness of control system types.
Thus, the ways that managers of di?erent kinds
try to control and coordinate the behaviour and
performance of organizational units are greatly
interdependent with the sorts of ®rms and orga-
nizations they are members of and the nature of
the political, ®nancial, labour and cultural systems
they are embedded within. This approach suggests
three general implications for the comparative
analysis of management control systems and
management accounting practices.
First, just as task uncertainty, employee skill
levels and environmental instability are widely
considered to a?ect the ecacy of bureaucratic,
output and normative control systems in the
organizational literature, so too do broader fea-
tures of organizations and their contexts. Indeed,
many of these ``contingencies'' are themselves
interdependent with societal institutions such as
trust and authority relations, labour systems and
the role of the state. For example, the manage-
ment of large organizations relying on technologi-
cally advanced and complex production processes
often requires considerable delegation of respon-
sibility to, and motivation of, highly trained tech-
nical sta? which may be dicult to achieve in
societies where trust in formal institutions and
relations is low. Not only, then, do control system
characteristics vary between organizations of dif-
ferent kinds and in di?erent sectors, as well as
between di?erent sorts of departments and hier-
archical levels, but they also vary signi®cantly
between types of ®rms and societal contexts such
that we would not expect any single mode of
ensuring control of economic activities to be gen-
erally applicable across such contexts. This is not
only because, say, the nature of ®nancial systems
a?ects the development of mutual dependence
between managers and employees and prevalent
standards of economic performance in an econ-
omy, but also because it in¯uences the sorts of
®rms that come to dominate that economy and the
ways they manage risks.
Anglo-Saxon capital market based ®nancial
systems, for instance, encourage a greater con-
centration on standardized ®nancial measures of
performance, diversi®ed, changeable ®rms and
limited managerial dependence on a single ®rm or
520 R. Whitley / Accounting, Organizations and Society 24 (1999) 507±524
sector, than does, say, the bank dominated one in
postwar Japan. Such standardized and abstract
measures in turn limit the development of highly
detailed prescriptions of unit behaviour, long
term, broadly based control systems and the
involvement of subordinates (Clark, 1979; Stewart
et al., 1994; Whitley, 1996, 1997). This is not to
say that in some ®rms and sectors these control
system characteristics may not occur but that
overall they will not become dominant ones in
such economies.
This interdependence implies that the develop-
ment and use of particular management accounting
procedures and practices by managerial groups in
di?erent ®rms, sectors and societies re¯ect sig-
ni®cant variations in the nature of these groups and
®rms, and their wider institutional contexts. An
important focus of management accounting
research, then, concerns how managerial elites of
various sorts in di?erent kinds of organizations,
®rms and societies come to rely on particular man-
agement accounting practices, and the consequences
of these variations for managerial decision making
and outcomes.
Second, the implementation and operation of
management control systems re¯ect a variety of
interests and capabilities which often con¯ict and
change over time. Comparative studies of the
machine tool industry in Britain, France, and
Germany have shown that the ways in which
managers control work processes are closely con-
nected to: (a) product market strategies and chan-
ges in these, (b) how they perceive their interests in
contrast to those of the skilled workforce, and, (c)
their ability to mobilise resources to achieve what
they want in the face of union, and sometimes
state, resistance (Sorge & Maurice, 1983; Piore
& Sabel, 1984; Herrigel 1994). As Lazonick
(1990, 1991) has suggested, the managerial desire
to remove skills from the shop¯oor to managers
and engineers in the U.S.A. was signi®cantly
in¯uenced by the ability of craft workers to con-
trol output and thereby limit the pro®tability of
the new technologies of mass production. Tight
bureaucratic control was thus a response to
adversarial labour relations where strong craft
unions preventedÐor threatened to preventÐthe
introduction of new production processes.
Elsewhere, and especially in parts of continental
Europe, ®rms were not so intent upon replacing
manual skills with capital intensive machinery and
craft workers were able to continue to play a sig-
ni®cant part in controlling work processes, partly
because of their relatively low wages and, in Brit-
ain at least, partly because of the unwillingness of
many employers to invest in the personnel and
procedures which would enable them to take direct
control of work processes. Control system char-
acteristics thus re¯ect competition and con¯ict
between various interest groups in the workplace,
and change when their nature and power alter.
This emphasises the contestable and contextual
nature of management accounting practices which
re¯ect and help to reproduce this competition and
con¯ict between variously constituted and power-
ful interest groups. The sorts of accounting proce-
dures and practices that are developed, and how
they are applied and interpreted, in di?erent cir-
cumstances are a?ected by the composition, prio-
rities and resources of owners, managers, and
labour representatives, and by the prevalent rules
of the game governing their competition and
cooperation. Changes in the nature of these
groups and their institutional contexts can be
expected to result in changes in accounting sys-
tems and how the information they generate is
used by various groups and linked to organiza-
tional change, as can be seen in the processes of
organizational restructuring in many of the former
state socialist societies of Eastern Europe (see, for
example, Clark & Soulsby, 1995; Konecki & Kul-
pinska, 1995; Nilsson, 1996).
Third, in seeking to understand whyÐor to
what extentÐcontrol systems vary across socie-
ties, it is important to explore the nature of the
processes connecting societal institutions, the con-
stitution and behaviour of interest groups, and the
sorts of ®rms that coordinate economic activities.
To search for direct correlations between, say,
``cultural'' predispositions as identi®ed by survey
questions and organizational control system char-
acteristics in isolation from these mediating and
interconnected phenomena is to risk drawing
rather super®cial and misleading conclusions. In
particular, it is important to be aware how speci®c
circumstances can a?ect the sorts of tendencies
R. Whitley / Accounting, Organizations and Society 24 (1999) 507±524 521
suggested in Table 4. In any particular organiza-
tion, sector or society, their use is mediated by
other factors which can severely inhibit their gen-
eral occurrence.
Predominantly paternalist authority patterns in
postwar Japan, for example, have not prevented
some involvement of middle managers in decision
making and of manual workers in problem solving
in many large Japanese ®rms, because of much
greater systemic trust in Japanese society than in,
say, Korea or Taiwan, and considerable mutual
dependence between employers and core employ-
ees through the long term employment system and
``inecient'' external labour markets (Rohlen,
1974; Clark, 1979; Dore, 1986; Koike, 1987; Aoki,
1988). This employment system itself is not, of
course, a permanent feature of the Japanese
industrial system but re¯ected particular manual
skill shortages, union structures and pressures,
and ®rm characteristics during the high growth
period (Murakami & Rohlen, 1992; Odagiri,
1992). It was easier to institutionalize in postwar
Japan because of the strong conventions govern-
ing obligations and cooperation within and
between organizations (Iwata, 1992), but these
relatively long lived features of Japanese society
were not sucient to bring it about in di?erent
socio±economic circumstances. Institutionalized
management control systems, then, are the out-
comes of speci®c combinations of processes and
in¯uences which empirically often generate con-
tradictory pressures. Accordingly, the compara-
tive analysis of such systems requires detailed
understanding of how the phenomena discussed
in this paper are interrelated in particular
®rms, industries and societies, and how they are
changing.
This sort of understanding is unlikely to be
achieved through simple correlational analysis of
responses to questionnaires which ignores the
speci®c context in which control systems operate
and the actions of participants. It involves
demonstrating how broad institutional features of
societies and cultures are linked to the speci®c
nature of ®rms and managerial and other groups
such that they develop particular logics and
rationalities which encourage the use of some con-
trol systems and accounting practices in preference
to others. It also involves studying the processes
by which, say, particular budgetary systems are
used and interpreted in practice, rather than simply
recording managers' responses to questions about
which procedures they follow. The comparative
analysis of management control and accounting
should, then, be based more upon ethnographic
and ®eldwork research, as well as detailed knowl-
edge of particular societies and organizations than
upon on postal surveys and descriptive statistical
analysis.
References
Aoki, M. (1988). Information, incentives, and bargaining in
the Japanese economy. Cambridge: Cambridge University
Press.
Bailes, J. C., & Assada, T. (1991). Empirical di?erences
between Japanese and American budget and performance
evaluation systems. International Journal of Accounting, 26,
131±142.
Beetham, D. (1991). The legitimation of power. London:
Macmillan.
Botti, H. (1995). Misunderstandings: A Japanese transplant in
Italy strives for lean production. Organization, 2, 55±86.
Boyer, R. (1991). Capital labor relation and wage formation:
continuities and changes of national trajectories. In T.
Mizoguchi (Ed.), Making economies more ecient and more
equitable, pp. 297±340. Tokyo: Kinokuniya.
Boyer, R. (1994). Do labour institutions matter for economic
development? a `regulation' approach for the OECD and
Latin America, with an extension to Asia. In G. Rodgers
(Ed.) Workers, institutions and economic growth in Asia (pp.
25±112), Geneva: ILO.
Boyer, R. (1996). State and Market. In R. Boyer and D.
Drache (Eds). States against markets, London: Routledge.
Campbell, J., Hollingsworth, R., & Lindbergh, L. (Eds) (1991).
Governance of the American economy. Cambridge: Cam-
bridge University Press.
Child, J., Fores, M., Glover, I., & Lawrence, P. (1983). A price
to pay? Professionalism in work organisation in Britain and
West Germany. Sociology, 17, 63±78.
Chow, C. W., Kato, Y., & Merchant, K. A. (1996). The use of
organisational controls and their e?ects on data manipula-
tion and management myopia. Accounting, Organizations and
Society, 21, 175±192.
Chow, C. W., Kato, Y., & Shields, M. D. (1994). National
culture and the preference for management controls.
Accounting, Organizations and Society, 19, 381±400.
Chow, C. W., Shields, M. D., & Chan, Y. K. (1991). The e?ects
of management controls and national culture on manu-
facturing performance. Accounting, Organizations and
Society, 16, 209±226.
522 R. Whitley / Accounting, Organizations and Society 24 (1999) 507±524
Clark, E., & Soulsby, A. (1995). Transforming former state
enterprises in the Czech Republic. Organization Studies, 16,
215±242.
Clark, R. (1979). The Japanese company. New Haven: Yale
University Press.
Cox, A. (1986). State, ®nance and industry in comparative
perspective. In A. Cox (Ed.), State, ®nance and industry.
Brighton: Wheatsheaf.
Deyo, F. C. (1989). Beneath the miracle. labour subordination in
the new asian industrialism. Berkeley: University of California
Press.
Dore, R. P. (1973). British factoryÐJapanese factory. London:
Allen and Unwin.
Dore, R. P. (1986). Flexible rigidities. Stanford: Stanford Uni-
versity Press.
Fields, K. (1995). Enterprise and the state in Korea and Taiwan.
Ithaca, NY: Cornell University Press.
Fligstein, N. (1990). The transformation of corporate control.
Cambridge, MA: Harvard University Press.
Fukuyama, F. (1995). Trust: the social virtues and the creation
of prosperity. London: Hamish Hamilton.
Gerlach, M. (1992). Alliance capitalism. Berkeley, California:
University of California Press.
Gold, T. B. (1986). State and society in the Taiwan miracle.
Armonk, NY: M. E. Sharpe.
Guillen, M. F. (1994). Models of management. work, authority
and organization in a comparative perspective. Chicago, IL:
University of Chicago Press.
Hamilton, G. (Ed.) (1991). Business networks and economic
development in east and southeast Asia. Hong Kong: Centre
of Asian Studies, University of Hong Kong.
Herrigel, G. (1993). Large ®rms, small ®rms and the govern-
ance of ¯exible specialisation. In B. Kogut (Ed.), Country
competitiveness. Oxford: Oxford University Press.
Herrigel, G. (1994). Industry as a form of order. In R. Hol-
lingsworth et al. (Eds). Governing capitalist economies.
Oxford: Oxford University Press.
Hollingsworth, R. (1991). The logic of coordinating American
manufacturing sectors. In J. L. Campbell, J. R. Hollings-
worth and L. Lindberg (Eds), Governance of the American
economy. Cambridge: Cambridge University Press.
Horowitz, J. (1980). Top management control in Europe. Lon-
don: Macmillan.
Inohara, H. (1990). Human resource development in Japanese
companies. Tokyo: APO.
Iwata, R. (1992). The Japanese enterprise as a uni®ed body of
employees. In S. Kumon and H. Rosovsky (Eds), The poli-
tical economy of Japan 3: cultural and social dynamics. Stan-
ford: Stanford University Press.
Janelli, R. L. (1993). Making capitalism. the social and cultural
construction of a South Korean conglomerate. Stanford:
Stanford University Press.
Jones, L. & Sakong, I. (1980). Government, business and entre-
preneurship in economic development: the Korean case. Cam-
bridge, MA: Harvard University Press.
Kim, C. S. (1992). The culture of Korean industry. Tucson:
University of Arizona Press.
Koike, K. (1987). The politics of economic management. In
K. Yamamura and Y. Yasuba (Eds), The political economy
of Japan I: the domestic transformation. Stanford, CA:
Stanford University Press.
Konecki, K. & Kulpinska, J. (1995). Enterprise transformation
and the rede®nition of organisational values in Poland. In E.
Dittrich, G. Schmidt and R. Whitley (Eds), Industrial trans-
formation in Europe. London: Sage.
Kristensen, P. H. (1992). Strategies against structure: institutions
and economic organisation in Denmark. In R. D. Whitley
(Ed.). European business systems: London: Sage Publications.
Kristensen, P. H. (1994). Strategies in a volatile world. Econ-
omy and Society, 23, pp.305±34.
Kristensen, P. H. (1997). National systems of governance and
managerial strategies in the evolution of work systems: Brit-
ain, Germany and Denmark compared. In R. D. Whitley
and P. H. Kristensen (Eds), Governance at work. the social
regulation of economic relations in Europe. Oxford: Oxford
University Press
Lane, C. (1989). Management and labour in Europe. Aldershot:
Edward Elgar.
Lane, C. (1992). European business systems: Britain and Ger-
many compared. In R. Whitley (Ed.), European business sys-
tems. London: Sage.
Lane, C, & Bachmann, R. (1996). The social constitution of
trust. Organization Studies, 17, 365±396.
Langlois, R. N. & Robertson, P. L. (1995). Firms, markets and
economic change. London: Routledge.
Lazonick, W. (1990). Competitive advantage on the shop ¯oor.
Cambridge, MA: Harvard University Press.
Lazonick, W. (1991). Business organization and the myth of the
market economy. Cambridge: Cambridge University Press.
Levy, B. (1991). Transaction costs, the size of ®rms and indus-
trial policy: lessons from a comparative case study of the
footwear industry in Korea and Taiwan. Journal of Develop-
ment Economics, 34, 151±178.
Lilja, K. (1997). Bargaining for the future: the changing habitus
of the shop steward system in the pulp and paper mills of
Finland. In R. Whitley and P. H. Kristensen (Eds), Govern-
ance at work: the social regulation of economic relations.
Oxford: Oxford University Press.
Luhmann, N. (1988). Familiarity, con®dence, trust: problems
and alternatives. In D. Gambetta (Ed.), Trust, Oxford:
Blackwell.
Maurice, M., Sorge, A., & Warner, M. (1980). Societal di?er-
ences in organising manufacturing units. Organisation Stu-
dies, 1, 59±86.
Maurice, M., Sellier, F., & Silvestre, J. J. (1986). The social foun-
dations of industrial power. Boston: MIT Press.
Merchant, K. A., Chow, C. W., & Wu, A. (1995). Measure-
ment, evaluation and reward of pro®t center managers: a
cross-cultural study. Accounting, Organizations and Society,
20, 619±638.
Murakami, Y., & Rohlen, T. P. (1992). Social-exchange aspects
of the Japanese political economy. In S. Kumon and H.
Rosovsky (Eds), The political economy of Japan 3: cultural
and social dynamics. Standford: Stanford University Press.
R. Whitley / Accounting, Organizations and Society 24 (1999) 507±524 523
Nilsson, K. (1996). Practice, myths and theories for change: the
reconstruction of an East German organisation. Organiza-
tion Studies, 17, 291±310.
North, D. C. (1990). Institutions, institutional change and eco-
nomic performance. Cambridge: Cambridge University Press.
O'Connor, N. G. (1995). The in¯uence of organizational cul-
ture on the usefulness of budget participation by singapore-
chinese managers. Accounting, Organizations and Society, 20,
383±403.
Odagiri, H. (1992). Growth through competition, competition
through growth. Oxford: Oxford University Press.
Orru, M., Hamilton, G., & Suzuki, M. (1989). Patterns of inter-
®rm control in Japanese business. Organization Studies, 10,
549±571.
Otley, D., Broadbent, J., & Berry, A. (1995). Research in man-
agement control: an overview of its development. British
Journal of Management, 6. (Special Issue) S31±S44.
Piore, M., & Sabel, C. (1984). The second industrial divide. New
York: Basic Books.
Redding, G. R. (1990). The spirit of Chinese capitalism. Berlin:
de Gruyter.
Rohlen, T. P. (1974). For harmony and strength: Japanese white-
collar organisation in anthropological perspective. Berkeley:
University of California Press.
Sako, M. (1992). Prices, quality and trust. Cambridge: Cam-
bridge University Press.
Shiomi, H., and Wada, K. (Eds) (1996). Fordism transformed.
Oxford: Oxford University Press.
Silin, R. H. (1976). Leadership and values. the organisation of
large scale Taiwanese enterprises. Cambridge, MA: Harvard
University Press.
Sorge, A., and Maurice, M. (1993). The societal e?ect in the
strategies of French and West German machine tool manu-
facturers. In B. Kogut (Ed.) Country competitiveness.
Oxford: Oxford University Press.
Steers, R., Shin, Y. K., and Ungson, G. R. (1989). The chaebol.
New York: Harper and Row.
Stewart, R., Barsoux, J-L., Kieser, A., Ganter, H-D., &
Walgenbach, P. (1994). Managing in Britain and Germany.
London: Macmillan.
Tolliday, S., & Zeitlin, J. (1991). National models and interna-
tional variations in labour management and employer orga-
nization. In S. Tolliday and J. Zeitlin (Eds), The power to
manage. (pp. 273±343). London: Routledge.
Ueno, S., & Sekaran, U. (1991). The in¯uence of culture on
budget control practices in the U.S.A. and Japan. Journal of
International Business Studies, 23, 659±674.
Wade, R. (1990). Governing the market. Princeton, NJ: Prince-
ton University Press.
Westney, E. (1996). The Japanese business system: key features
and prospects for changes. Journal of Asian Business, 12, 21±
50.
Whitley, R. (1992a) Business systems in east Asia: ®rms, mar-
kets and societies. London: Sage.
Whitley, R. (1992b) Societies, ®rms and markets: the social
structuring of business systems. In R. Whitley (Ed.), Eur-
opean business systems. London: Sage.
Whitley, R. (1996). The social construction of economic actors:
institutions and types of ®rm in Europe and other market
economies. In R. Whitley and P. H. Kristensen (Eds), The
changing European ®rm. London: Routledge.
Whitley, R. (1997). The social regulation of work systems:
institutions, interest groups and varieties of work organisa-
tion in capitalist societies. In R. D. Whitley and P. H. Kris-
tensen (Eds), Governance at work: the social regulation of
economic relations in Europe. Oxford: Oxford University
Press.
Whitley, R., & Kristensen, P. H. (Eds) (1996). The changing
European ®rm. London: Routledge.
Woo, Y.-E. (1991). Race to the swift. New York: Columbia
University Press.
Zucker, L. (1986). Production of trust: institutional sources of
economic structure 1840±1920. Research in Organisational
Behaviour, 8, 53±111.
Zysman, J. (1983). Governments, markets and growth. Ithaca:
Cornell University Press.
524 R. Whitley / Accounting, Organizations and Society 24 (1999) 507±524

doc_469850006.pdf