10.1016@s0361 36820000041 6
10.1016@s0361 36820000041 6
10.1016@s0361 36820000041 6
www.elsevier.com/locate/aos
Abstract
This paper argues that the discipline of management control is likely to bene®t from theories that provide a more
cogent and comprehensive perspective to address the issue of control structure variety, and examines the potential of
Transaction Cost Economics (TCE) to inform such theories. Based on TCE, this paper proposes a framework that
explicates the link between various archetypical con®gurations of control devices and the activities they are expected to
control. In this framework, the nature of the organizational activities and the required contributions from organiza-
tional participants are de®ned along three dimensions: (1) the extent of programmability; (2) the degree of asset speci-
®city; and (3) the intensity of ex post information impactedness. These attributes are associated with distinctive control
problems that need to be dealt with. The control archetypes dier in their problem-solving ability, which makes them
appropriate for the governance of some contributions, but not for others. Moreover, they dier in cost, and the fra-
mework explains the alignment of a contribution with a control archetype by delineating the eciency properties of the
match. # 2001 Elsevier Science Ltd. All rights reserved.
Keywords: Management control theory; Transaction cost economics
that the appropriateness of MC structures is choice. TCE has proven quite successful at this in
aected by some set of circumstantial factors its own domain, and trying to transfer the insights
faced by the organization. But this approach is accumulated there to the domain of MC might
more a general idea than an articulated theory in well be worth the eort. Appearances may be
the sense that it has no a priori intuition of its own misleading, though, and Section 3 of this paper
as to what the pertinent factors are and as to their provides a more thorough examination of the
likely consequences. Of course, researchers within degree of correspondence between commonly held
that stream do have pretty clear notions about the perspectives on what MC is about and on the
variables and their eects, but these notions come direction in which to search for explanations on
from sources outside contingency theory per se. the one hand, and the theoretical angle of TCE on
These sources are diverse and tend to dier widely the other. This assessment allows a preliminary
across researchers, on account of which the con- consideration of TCE's conceptual relevance to
tingency stream is rather heterogeneous (cf. issues of MC, and helps to clarify the general nature
Chapman, 1997; Fisher, 1998; Lang®eld-Smith, and scope of TCE's potential contribution. The
1997; Otley, 1980). Moreover, contingency models second step is to move beyond mere consideration
tend to be partial, focusing on elements of control and involves a concrete demonstration of what
systems (e.g. budgeting systems) as opposed to TCE is able to oer. To this eect, the logic of
addressing directly the full con®guration of control TCE is asked to speak for itself in Section 4, which
devices. For these reasons, the contingency presents a tentative transaction cost theory of MC.
approach does not provide a self-sucient and Section 5 contains a summary and discussion of
fully ¯edged perspective on the issue of MC the argument. But ®rst, Section 2 provides an
structure variety. Agency theory suers from introduction to TCE's analytical toolbox and line
eectually similar weaknesses. Because agency of reasoning.
theorists are known to have a strong predilection
for formal mathematical modelling, they normally
choose to focus on a limited number of well-spe- 2. An introduction to TCE
ci®ed, highly stylized elements of MC systems (cf.
Merchant & Simons, 1986; Scapens, 1991), rather TCE studies organization from a comparative
than attempting to capture control in its entirety. point of view in which dierent institutional
They furthermore tend to con®ne themselves to arrangements are considered alternative ways to
single period, single agent models (Baiman, 1982, organize economic activity. Its central aim is to
1990), which may lack representational validity. explain why some transactions are more likely to
Therefore, their contribution to knowledge in the be executed within one form of organization,
®eld of MC Ð as that of their contingency theory whereas other transactions tend to be associated
counterparts Ð remains fragmented, whereas with dierent organizational modes. The main
more inclusive approaches are also needed to thrust of TCE's answer is that a speci®c institu-
understand MC. tional arrangement is chosen to govern a speci®c
This paper intends to contribute some insights transaction because that arrangement oers some
that may increase the theoretical coherence of the distinctive set of control devices Ð a set that cannot
study of MC structure variety. More speci®cally, it be replicated within alternative arrangements Ð
examines the potential of transaction cost eco- that is uniquely tailored Ð in a relative sense Ð to
nomics (TCE; Williamson, 1975, 1979, 1985, 1996) the control needs of that transaction. Thus, TCE
to provide a cogent, parsimonious perspective that submits that transactions dier in respect of the
is able to inform the study of MC. A prima facie, contractual problems to which they give rise, whilst
this seems a reasonable thing to do, because TCE organizational forms dier in their problem-solving
and MC share a common interest in under- ability, so that alignments between the two can be
standing purposive control, and both are com- explained by explicating the ecacy and eciency
mitted to the explanation of control structure of the match.
R.F. Spekle / Accounting, Organizations and Society 26 (2001) 419±441 421
or arbitration) when it deprives the enforcement Market control is to be expected for transac-
agency of the information necessary for fair con- tions that score low on asset speci®city. Absence
¯ict settlement. The role of frequency is that it of idiosyncrasy implies the availability of a large
exacerbates the contracting problems associated number of alternative contracting parties and low
with the other dimensions, adding to the pressure switching costs. The irresistible `invisible hand'
to ®nd a solution. takes care of contract enforcement and adaptation
to changing circumstances. Since parties have
2.2. Contracting solutions access to alternative sources of supply and
demand, competition quite eectively curbs any
Economic factors try to cope with these problems opportunistic inclinations, and it does so at low
by means of organization, i.e. by adopting appro- costs. Rising asset speci®city, however, impedes
priate organizational arrangements to govern their the smooth functioning of the market, and alter-
transactions. At a generic level, TCE de®nes three native governance devices are now needed to protect
distinct modes of governance: (1) markets; (2) the transaction against opportunistic breach of
hybrids; and (3) hierarchies or internalization. contract and uncooperative gap-®lling behaviour.
These alternative governance structures dier in The hybrid mode of governance may oer these.
the control mechanisms they employ to safeguard Compared with the market, hybrid governance
contract execution and to achieve successful structures provide additional transaction-speci®c
adaptation. Market governance derives control from safeguards such as hostage arrangements and spe-
free competition. The hybrid form of governance is cialized dispute settlement institutions (Williamson,
typically based on explicit, long-term contracts in 1996, pp. 120±144). These safeguards serve to
conjunction with additional safeguards to assure advance compliance to the provisions of the con-
compliance. Hierarchical governance attains control tract. Therefore, their application requires at least
primarily by means of authority, internal incentive some previously agreed to, contractually anchored
structures, and monitoring. notion as to what constitutes adequate perfor-
But the structural options also dier in respect mance. In conditions of substantial uncertainty,
of costs. TCE's main theme is that transactions Ð the minimum level of contractual detail cannot
which dier in their attributes Ð are aligned with always be provided. Moreover, these safeguards
governance structures Ð which dier in their costs are usually less than perfect, which may be pro-
and competencies Ð in a discriminating, econo- blematic when asset speci®city rises. In special
mizing way. Transaction cost eciency, thus, is conditions, these problems can be coped with
pertinent to the explanation of the match between within the hybrid form by invoking mechanisms
governance structures and transactions. It is, that secure goal congruence and that mitigate
however, not fully deterministic: ``if economic information asymmetry. But usually, governance
organization is formidably complex, which it is, structures that more reliably secure adaptive,
and if economic agents are subject to very real cooperative attitudes and actions when it comes to
cognitive limits, which they are, then failures of ®lling the contractual gaps are needed, and hier-
alignment will occur routinely'' (Williamson, 1996, p. archical governance may be appropriate.
311). Nevertheless, a general tendency towards e- The hierarchical solution is to evade con¯icts of
ciency is presumed to exist and to drive the con®g- interest between contracting parties. Within the
uration of instruments of control. Neither is eciency hierarchy, individual compensation is typically not
to be interpreted in some absolute sense. In TCE, based solely on the direct, isolated ®nancial out-
eciency is a relative concept which takes the form of comes of internal transactions. The increasing
a `remediableness criterion'. This criterion holds that incidence of explicit performance-related bonus
an existing con®guration of governance devices is plans notwithstanding, the hierarchy tends to
(refutably) ecient when no feasible alternative can emphasize ®xed compensation as its primary reward
be described and implemented with expected net system, linking promotion to contributions made to
gains (Williamson, 1996, 1999a,b). the accomplishment of usually underspeci®ed
R.F. Spekle / Accounting, Organizations and Society 26 (2001) 419±441 423
organizational goals. This practice supports coop- domain of MC. To avoid an all too personal bias,
eration, and establishes quite a large zone of this section will start from an examination of some
indierence in which choices and changes can be broad de®nitions of the ®eld of MC. Presuming
implemented by simple managerial ®at. The reliance that such de®nitions adequately re¯ect common
on managerial discretion oers signi®cant decision perceptions about the purposes of MC, they may
making ¯exibility and permits sequential adapta- provide some insight into the criteria and desiderata
tion. It furthermore allows incorporation of less that are implicit in current MC-thinking and that
explicable aspects of performance in performance may be used to explore the conceptual suitability
appraisal, thus avoiding myopic pursuit of pre- of TCE to inform the study of MC.
de®ned but incomplete standards. However, there
is a price involved here as well. Managerial coor- 3.1. What is MC about anyway?
dination, the required administrative apparatus
and information systems are imperfect and hardly MC has been de®ned in numerous dierent
costless. The almost innate limitations of accounting ways. Nevertheless, the many de®nitions that are
information systems need no elaboration here, and proposed in the literature do share a lot of common
the all too familiar bureaucratic inertia and politi- ground. Although MC has also been studied in a
cizing of decision making are just a few examples political arena, e.g. as a tool to conceal or legit-
to illustrate that hierarchical governance is no imize the exercise of power, as a means to exploit
panacea. the working class, or as an instrument to construct
an image of rationality that better ®ts constituents'
expectations (see for instance Covaleski, Dirsmith
3. TCE and MC: some preliminary considerations & Samuel, 1996, for an overview of this literature),
most appear to describe MC as a purposive process
Although TCE has some history in management or set of devices and mechanisms, that through its
accounting and control (for instance Johnson, in¯uence on the behaviour of actors within an
1983; Seal, 1993; Spicer & Ballew, 1983; Swieringa organization, intends to contribute to the achieve-
& Waterhouse, 1982; Tiessen & Waterhouse, 1983; ment of some pervasive objectives of that organi-
Van der Meer-Kooistra, 1994; Van der Meer- zation. Thus Flamholtz de®nes an organizational
Kooistra & Vosselman, 2000), Baiman's observa- control system as ``a set of mechanisms which are
tion that TCE ``has not, as yet, had a signi®cant designed to increase the probability that people
eect on the direction of managerial accounting will behave in ways that lead to the attainment of
research'' (p. 346) is as true today as it was a decade organizational objectives'' (quoted in Merchant,
ago when he wrote these words, and is also true 1985, p. 5). Ansari's de®nition is concordant: ``a
for the wider domain of MC. The reason for this control system [. . .] may be simply described as those
lack of impact is not entirely clear, but it is con- organizational arrangements and actions designed to
ceivable that some fundamental discrepancy facilitate its members to achieve higher performance
between the demands of the domain that MC with least unintended consequences'' (Ansari, 1977,
purports to cover on the one hand, and the nature p. 102; italics in original). Yet another de®nition
and scope of the explanatory apparatus of TCE sees MC as ``the process of guiding organizations
on the other explains TCE's virtual absence from into viable patterns of activity in a changing
MC-thinking. This current section attempts to environment'', which statement the authors take
examine the degree of correspondence Ð or con¯ict, to imply that control is about in¯uencing ``the
as the case may be Ð between MC and TCE. Such behaviour of [. . .] organizational participants so
an evaluation of basic compatibility requires an that some overall organizational goals are
explication of what MC-theory should ideally be achieved'' (Berry, Broadbent, & Otley, 1995, p. 4).
able to do, and an identi®cation of the general And after a review of some of the relevant literature,
properties of aspiring core theories of MC that are Merchant and Simons (1986) note that control
required to enable sucient coverage of the de®nitions generally include two key concepts: a
424 R.F. Spekle / Accounting, Organizations and Society 26 (2001) 419±441
focus on the behaviour of organizational partici- largely to issues at the level of the generic forms, this
pants, and a concern with the eect of this behaviour emphasis is more coincidental than fundamental.
on organizational outcomes. These de®nitions are TCE's explanatory structure is highly micro-ana-
presumably quite useful in that they seem to lytical in orientation, its unit of analysis being the
encapsulate a signi®cant part of the work that is (individual) transaction and its focus being on the
being done in the name of MC, and will be used as problems of governance that are associated with
a reference point in the preliminary assessment of that particular transaction. Now it is true that the
the suitability of TCE to guide research in MC. problem-solving devices (the governance struc-
tures) have been formulated at a markedly high
3.2. The level of analysis level of aggregation, but that is mainly because the
originally envisioned and initially attempted appli-
Apparently, MC is regarded as a functional cations happened to demand that level. The
system or process, its purpose being the further- micro-level at which TCE's variables are located
ance of some organizational goal or goals. This permits a more detailed study of governance than
view suggests that MC is an instrument of organi- is usual in TCE-based studies, and on this
zation and that it must be understood within the account, there is no a priori obstacle to TCE's
organizational context in which it operates and of application to issues of MC.
which it is part (cf. Birnberg, Turopolec & Young,
1983; Hopwood, 1978, 1983; Lowe & Chua, 1983; 3.3. A behavioural focus
Merchant & Simons, 1986; Otley, 1984). Under-
standing MC thus presupposes an understanding There is broad consensus that MC works
of organization, and basic theories in MC are through its in¯uence on people. As a consequence,
likely to share much of their explanatory apparatus not any theory at the organizational level will do
with organization theory. At least, they share a equally well, and to succeed as a basis on which to
common level of analysis Ð that of the organiza- build MC-theory, the candidate theory must
tion and its substantive parts Ð , albeit that MC- incorporate considerations of a behavioural kind
theory may require a higher level of resolution: in its explanation. A prima facie, TCE presents no
since MC concentrates on a particular aspect of problems here, for the notions of bounded ration-
organization, more detail may be necessary. ality and opportunism are undeniably behavioural
It is quite possible that the lack of impact of indeed. But then again, it is not entirely clear what
TCE on MC-thinking is caused by concerns the academic MC-community in general would
regarding the level of analysis. After all, TCE has consider to be sucient behavioural content. After
its source in the domain of microeconomics, and all, a model of man based solely on these notions
has been used primarily to explain the trade-o is not exactly descriptively rich. It may even be
between the generic modes of governance. MC on criticized as derisively incomplete. On the other
the other hand is more interested in trade-os hand, however, it may be argued that fullness of
within just one of the generic modes: the hierarchy. behavioural detail is not a relevant yardstick by
But if this really is the reason, then it is mistaken. which to measure candidate theories. MC does not
TCE acknowledges the view that the distinct generic purport nor need to portray man in his entirety,
structures are heterogeneous categories that sub- but may settle for an instrumental understanding
sume a variety of dierent con®gurations of con- of how human behaviour aects control problems
trol devices Ð that is why they are called generic. and solutions. If one accepts this position, there is
Yet, TCE suggests that these diverse control not much gain in evaluating the assumptions as if
arrangements can meaningfully be addressed as they were a theory of human behaviour in their
variations on the transaction cost theme, using its own right Ð for obviously, they are not Ð , but
general analytical apparatus. And there is no they should be judged by their implications for the
inherent reason why this suggestion would not be domain of MC and predictive value in that ®eld.
valid. Although its application has been con®ned Both bounded rationality and opportunism are in
R.F. Spekle / Accounting, Organizations and Society 26 (2001) 419±441 425
fact the source of many of the refutable implica- pervasive objectives that give direction and coher-
tions of TCE, and are likely to be similarly pro- ence to the combined eorts of the members of the
ductive when applied in the context of MC. But organization, and need not be concerned with the
whether suciently full predictive coverage of the source and nature of these objectives. Presuming
relation between human behaviour and control is the existence of such objectives, MC may treat
included in the bargain is an empirical matter that these as a given and concentrate on the question as
cannot be settled here. to how MC structures and practices help to
accomplish these goals. This is obviously a some-
3.4. Goals and eectiveness what blinkered approach, but awaiting a more
balanced point of view that integrates the political
The generally subscribed to view of MC as a and conventional perspectives, some abstraction is
means to assist achievement of organizational unavoidable.
goals implies that ultimately, explaining MC must But then, concentrating one's eorts on the
come down to demonstrating the actual contribu- analysis of how MC supports goal attainment has
tion of observed MC practices to the attainment problems of its own, because the actual goals may
of these goals. MC is about organizational eec- be quite dicult to identify. The actual goals do
tiveness, whence theories of MC require some not necessarily coincide with the ocial goals, for
form of speci®cation of the objectives of the orga- the readily observable, ocially stated objectives
nization and must be able to relate MC to the (e.g. those expressed in mission statements and
attainment of these objectives. This, however, is declarations of corporate policy) tend to be cast in
problematic (Berry et al., 1995; Lowe & Chua, a language that is perhaps inspirational but too
1983; Lowe & Puxty, 1989; Otley, 1983, 1999). vague and insuciently operational to re¯ect what
One problem that arises here is the convoluted really moulds the organization's accomplishments
issue as to whether organizations can in fact have and eorts (cf. Lowe & Chua, 1983), Then, the
goals or that having goals is the province of the true operative objectives need to be inferred from
individuals comprising the organization (Cyert & what goes on in the organization.
March, 1963). Also, there is the issue as to the One popular way to deal with this diculty has
source of the objectives, involving questions as to been to assume that organizational goals Ð what-
the process by which they are arrived at, by whom ever they are and wherever they come from Ð are
they are in¯uenced, and by what means this in¯u- translated into fairly concrete and explicit strategies
ence is exerted (Berry at al., 1995; Lowe & Chua, that are to guide organizational behaviour
1983; Otley, 1980). Such issues are absolutely cru- towards goal attainment. How this translation is
cial if one chooses to study MC in relation to being made is an issue that is left to the domain of
power, politics, ideology, and con¯ict, and in its strategic planning and is considered to be outside
role in the construction of perceived societal or MC's territory. This approach, epitomized by the
organizational reality1. One could, however, argue in¯uential work of Anthony (1965, 1988), con®nes
that for the purposes of MC in its more conven- the function of MC-systems and processes to
tional conception, i.e. being about the furtherance ensuring that the formulated strategies are being
of organizational goals, these issues are not neces- implemented as planned. This move, however, is
sarily part of the core problem. In that view, MC unsatisfactory for it ignores that strategies, rather
may rest content with the identi®cation of some than invariably being the result of explicit and
tractable prior decision making, frequently
1
This is not the place to discuss these more `critical per- emerge and evolve during the course of business
spectives'. The richness and diversity of these perspectives (Mintzberg & Waters, 1985). It also neglects the
become quite clear from the paper collections edited by Chua,
constituent role of MC in crafting strategy
Lowe and Puxty (1989) and Cooper and Hopper (1990). Other
useful references would include for instance Carruthers (1995), (Simons, 1990, 1995). The Anthony-approach,
Covaleski and Aiken (1986), and Covaleski, Dirsmith and therefore, is bound to miss much of what drives
Samuel (1996). the organization.
426 R.F. Spekle / Accounting, Organizations and Society 26 (2001) 419±441
Essentially, there seem to be two dierent lines According to TCE, (relative) economic eciency is a
of attack to deal with this issue. One is to attempt genuine survival condition, and its pursuit Ð i.e. the
to identify the general and fundamental objectives avoidance of maladaptation and waste Ð is crucial,
that are shared by all organizations alike Ð as for on account of which TCE asserts the urge to econo-
instance long-run survival (Lowe & Chua, 1983) Ð, mize on transaction cost to be a crucial determinant
and then to focus on the contribution of MC in of control arrangements. But it simultaneously con-
realizing these universal objectives. The other is to centrates the analysis on the actual transactions that
work from the empirically identi®ed patterns of occur within the organization and on those that cross
actions, decisions, and beliefs that characterize the its boundaries, and con®nes itself to the question
organization and that shape its eorts and func- whether these are being controlled eciently.
tioning. These patterns of ``organizational order'' Such an analysis is obviously not exhaustive.
(Dermer, 1988) may be taken as the organization's For one, the analysis ignores objectives of a non-
apparent operationalization of the actual objec- economizing kind. This is `negligence by design' of
tives, and MC and its contribution to eectiveness which TCE is fully aware: TCE holds economizing
may be studied in relation to these patterns. The on transaction cost to be the main case and not
latter approach seems the more promising, the only case (Williamson, 1996, 1999b; William-
because the former is so broad and indeterminate son & Ouchi, 1981). But inasmuch as economizing
that it is unlikely to render any discriminating is in fact a main case factor in control structure
results. It bears notice, however, that taking the choice Ð and this is an empirical question that
second route implies that MC can no longer cannot be answered a priori Ð , starting from
rightfully claim that it is concerned with organi- there is entirely defensible, if only as a matter of
zational eectiveness per se, because by working priority. The second blind spot is that the analysis
out of actual patterns of organizational behaviour, more or less takes for granted the existing set of
it takes the existing patterns for granted. The transactions, i.e. the apparent operationalization
focus of MC, then, shifts to more instrumental of the organization's goals and strategies. To be
eectiveness Ð is control eective in supporting sure, the TCE approach is able to incorporate
what the organization apparently tries to considerable contextual detail by studying the
achieve? Ð , and the higher level question as to the transactions in relation to the patterns of activ-
overall ecacy of the particular patterns in light ities, beliefs and intentions of which they are part
of the organization's ultimate objectives Ð is the in an eort to interpret the concrete transactions
organization trying to achieve the right things? Ð in terms of their role and meaning in the organi-
is neither asked nor answered, at least not in a zation's endeavours. The resulting insights enter
systematic, comprehensive way. But considering the analysis as elements of the contractual pro-
that this more humble aim leaves a lot to be dis- blem, for they either co-de®ne desired perfor-
covered, one may well be prepared to live with mance or in¯uence the characteristics of the
that. Be that as it may, no basic theory of MC can transactions that need to be governed. But the
avoid to explicate its criterion of eectiveness. approach ignores the question as to the overall
Speci®cation of the goals Ð fundamental, instru- eectiveness of these transactions in light of the
mental, or otherwise Ð is thus essential, and ultimate objectives of the organization. Referring
explication of the way in which MC assists to to the earlier discussion of this point, however,
accomplish these goals is required. one may argue that this problem is insoluble Ð at
TCE more or less combines the `universal' and least for now. This, then, would be `negligence by
the `instrumental' approaches. TCE almost profa- necessity' with which one has to learn to live.
nely submits that whatever the full range of
objectives of the organization, there always is the 3.5. Dierential eectiveness
pervasive and dominant need to work out the
alignment between transactions and governance Although explaining MC requires explication of
structures in respect of transaction cost eciency. how MC supports organizational eectiveness, a
R.F. Spekle / Accounting, Organizations and Society 26 (2001) 419±441 427
and (3) the intensity of post hoc information and information to decide in advance on the way
impactedness. The ®rst two of these variables are in which they are to be executed in order to
taken directly from TCE. The last one can also be achieve success, or activities for which the out-
traced to TCE, but is given a more prominent comes that may realistically be expected to result
place and is more fully worked out to increase the from them can be de®ned ex ante; and (2) non-
expressiveness of this variable and its eects on programmable activities, i.e. activities for which
control structure choice. These variables and their the organization lacks the a priori ability and
consequences are used to explain nine dierent experience to relate actions to outcomes. The
control archetypes, ®ve of which belong to the availability of norms and standards in the ®rst
hierarchical mode of governance (Tables 1 and 2). group permits a fairly comprehensive ex ante
Some of these (especially machine control and articulation of the characteristics of the contribu-
arm's length control) correspond to well-known tion that is required from the members of the
patterns of control from the MC literature. Others organization, and contracting for that contribution
(exploratory and boundary control) are relatively can be reasonably complete. Control, therefore, is
new as theoretical constructs, although I believe expected to be prescriptive or authoritative in
that their relevance to the ®eld of MC is easily nature, featuring rules of behaviour, speci®c
recognized. Exploratory control for instance instructions, and relatively rigid performance targets,
seems to capture much of Ð or at least relate to Ð and to be focused on assuring compliance to these
what has recently been reported in case studies pre-imposed norms. In the second group, in contrast,
addressing control in circumstances of ambiguity it is not possible to specify required contributions in
(see for instance Marginson, 1999; Mouritsen, advance. Due to the absence of ex ante standards,
1999, and Ð less directly Ð Vaivio, 1999a,b). And contracts must be of a general thrust nature,
boundary control is reminiscent of Simons' (1995) emphasizing a general commitment or sketching
use of that term. the broad con®nes within which performance
ought to ®t, rather than delineating a precisely
4.1. The eects of uncertainty: programmable speci®ed contribution.
versus non-programmable contributions
4.2. An interlude: low asset speci®city
In TCE, uncertainty is a condition that can arise
from many sources, including market dynamics, The nature of control is further de®ned by the
disturbances in the external environment, environ- degree of asset speci®city or idiosyncrasy of the
mental complexity, task uncertainty and complexity, activities. The level of idiosyncrasy aects the
and unfamiliarity. However, whatever the source, options which are available to secure or elicit
the eects are similar: transactions are not amen- contract congruent behaviour. Low asset speci®-
able to up front programming, and maintaining city implies that the desired contribution is of a
¯exibility to allow adaptation to events as they general purpose kind, not involving assets that are
unfold and to information as it accrues becomes tailored to the organization. Such contributions
imperative. This basic insight Ð which also has a are likely to be governed by the market mechanism.
long history in MC, albeit under dierent names In this situation of large numbers bidding, the
and in various guises2 Ð allows organizational market is quite able to curb opportunism and to
activity to be grouped in two broad categories: (1) secure adaptation when circumstances change;
programmable activities, i.e. activities for which both opportunism and non-adaptive behaviour
the organization possesses sucient knowledge would mean to lose customers and, therefore,
carry their own demise. Moreover, the market
2
mechanisms comes cheaply here, which makes it
Early references would include for instance Burns and
Stalker (1961) and Galbraith (1973). Some recent reviews of the
the ecient control choice. This prediction is
abundance of empirical work in this area can be found in independent of the degree of programmability or
Chapman (1997), Fisher (1995, 1998) and Hartmann (2000). uncertainty. Because adaptation is the almost
R.F. Spekle / Accounting, Organizations and Society 26 (2001) 419±441 429
automatic, irresistible result of a competitive process to engage in such games. Buttressed with mon-
and not the outcome of sovereign decision-making by itoring and a policy that rewards complaisance,
individual actors with their potentially opportunistic this structure may eectively assure compliance.
propensities, market governance can tolerate sig- This type of control strongly resembles the kind of
ni®cant uncertainty and does not require ex ante control the proponents of conventional responsi-
speci®cation of desired performance. The condition bility accounting had in mind when making their
of low asset speci®city thus suppresses the eect of recommendations. It is also closely related to the
uncertainty described in Section 4.1: whatever the `mechanistic organization' described by Burns and
degree of programmability, the `invisible hand' Stalker (1961) and the `machine bureaucracy'
takes care of control. Although a large number of portrayed by Mintzberg (1983). It features standar-
®rms appear to rely on market-like instruments dization and regulation of behaviour, codi®cation
for control purposes (e.g. market-based transfer of budget targets, detailed monitoring, systematic
prices or performance benchmarks) the Market measurement of performance on pre-de®ned
Control archetype in its pure form is, however, dimensions, and clearly identi®ed areas of
generally considered to be outside the domain of accountability, usually mirrored in the organiza-
MC, and is referred to solely for reasons of fullness tional structure. Its emphasis on programming,
of exposition. Whilst due attention will be given to progress monitoring, and correcting deviations
control archetypes that absorb market discipline from pre-set directions suggests the label Machine
in their hybrid or hierarchical structures, no further Control for this structure.
examination of pure market control will be The machine control archetype can be re®ned by
attempted here. distinguishing Action Oriented and Result Oriented
machine control types. In the action oriented
4.3. Programmable activities and high asset approach, control is predominantly achieved via
speci®city: Machine Control codi®cation of actions and supervising observance
of the rules and instructions, whereas control of
Rising asset speci®city implies a decreasing the result oriented kind hinges primarily on target-
number of potentially competent contributors and setting, accountability, and reward structures that
an increasing dependency on those few contributors serve to encourage target-directed behaviour. This
that are in fact available. And inasmuch as the distinction has been dealt with quite extensively in
contributor is to invest signi®cantly in specialized the literature Ð see for instance Merchant's
assets (including knowledge and skills) as a pre- results controls and action accountability controls
requisite for his or her contribution, a reverse (Merchant, 1982, 1985), and Ouchi's behaviour
dependency relation builds up as well. The resultant control versus output control (Ouchi, 1977) Ð
lock-in eect entails the need for special devices to and need no ampli®cation here, except for the
alleviate opportunism. eciency properties of the alternatives.
In the group of programmable activities, where In many instances, there will be no real choice
control takes a prescriptive orientation, the between action oriented control and the result-
emphasis will be on compliance to the prede®ned oriented approach, simply because the available
norms and standards. Given a high degree of asset information enables the one and not the other
speci®city, contracts Ð complete as they may be Ð (Merchant, 1982, 1985). Then, straightforward
are not easily enforceable in a legalistic or for- feasibility considerations will be decisive. But
malistic sense, for the (mutual) dependency allows when both approaches are feasible, result control
substantial leeway for strategic ploys, if only will usually reign for it tends to require less ela-
because high asset speci®city tends to involve low borate structuring Ð thus relieving the pressure
observability to outside enforcement agencies. The on bounded rationality Ð , is likely to demand less
con¯ict-evading hierarchical solution of easing the higher level involvement, and is more supportive of
link between individual compensation and the adaptation. The latter aspect is important when Ð
outcomes of the contribution mitigates the incentive low uncertainty notwithstanding Ð there may still
430 R.F. Spekle / Accounting, Organizations and Society 26 (2001) 419±441
be some unanticipated disturbances or opportunities arrangements are contracts that specify some form
demanding a ¯exible response. The result control of penalty payment to the compliant party to
variant may rely on a performance-dependent compensate for any non-salvageable losses incurred
reward system to provide the incentive to elicit because of contract defection by the other party,
that response, whereas the action oriented alter- or the provision that require a party to commit
native has no such option and needs to revert to resources that are similarly aected by the outcome
time-consuming hierarchical rede®nition of required of the transaction to balance speci®c investments
behaviour. made by the other party to the exchange. In addi-
tion to such hostages, specialized arbitration may
4.4. Programmability and moderate asset be installed to settle any disputes that may arise
speci®city: Arm's Length Control during contract execution.
If, however, ecient hostage arrangements cannot
Between the extremes of high and low asset be drafted, hierarchical arm's length control may
speci®city (fostering machine control and market supplant the hybrid form. Also, the hierarchical
control respectively) are the activities of moderate variant may arise when (full) outsourcing (the
idiosyncrasy. This is the domain of combinatory hybrid solution) is considered premature or
forms of control, in which residual market dis- otherwise undesirable. The organization may for
cipline is joined with elements of administrative instance want to retain some in-house production
origins (the hierarchical type) or with hostage as a cost-eective disciplinary device in its relations
arrangements (the hybrid type) to safeguard trans- with outside contractors and as a credible threat in
actions. These control structures are ecient when its negotiations with these suppliers. Or it may
competition is too thin to bank on the ecacy of want to preserve its access to the technology
market control, whilst the activities are not su- involved in the activity for strategic reasons, as
ciently essential to warrant fully ¯edged machine would be the case when the organization regards a
control with its bureaucratic burden. Typical for technology that is currently hardly vital, to embody
these control solutions is that the contributor a valuable option on future growth. Hierarchical
retains signi®cant autonomy, whence these con- arm's length control relies to a signi®cant extent on
®gurations of control can be referred to as Arm's market mechanisms, but provides additional and
Length Control. Consider the hybrid variant ®rst. more active means to in¯uence performance. It more
Hybrid governance structures are de®ned as or less imports market discipline in the hierarchy,
``long term contractual relations that preserve treating the department or division as independent,
autonomy but provide added transaction-speci®c assessing its record relative to outside competitors'
safeguards compared with the market'' (Williamson, performance, and (probably) tying rewards to
1996, p. 378). Thus, they refer to contracting performance, whilst maintaining recourse to the
between autonomous market parties, i.e. to out- rich repertoire of managerial intervention when
sourcing relationships. The long-term nature of performance drifts out of line.
the contract is indicated to enable compensation
for durable transaction-speci®c investments. The 4.5. Non-programmable activities: a note on post
added safeguards frequently include reliance on hoc information impactedness
hostage arrangements to correct asymmetric
stakes in the contract. In such arrangements, parties The in¯uence of idiosyncrasy on the governance
to the contract are required to make investments of non-programmable activities is akin to the
or to transfer valuable assets, the full cost of which eects described for the programmable class: low
can only be recovered in case of successful con- asset speci®city favours market control, moderate
tract execution, thus curtailing the potential gains levels of idiosyncrasy indicate hybrid or hierarchical
from opportunistic defection and compensating control, and highly speci®c transactions request
for the loss of market discipline that results from hierarchical governance. A closer examination will
small numbers bidding. Examples of these hostage follow shortly. But ®rst, another variable needs to
R.F. Spekle / Accounting, Organizations and Society 26 (2001) 419±441 431
be addressed: the level of post hoc information the impossibility to protect the information from
impactedness, i.e. the extent to which the organi- opportunistic manipulation by the sender at
zation is able to observe and to assess perceptively acceptable cost. In the case of irremediable
the true quality of actually delivered contribu- impactedness, the availability of control mechanisms
tions. The relevance of this variable is con®ned to that may actively ensure desired contributions is
the category of non-programmable activities; in severely limited, and the aim of control shifts to
the case of the more programmable ones the the prevention of undesired actions or outcomes.
required information must by de®nition be avail- As suggested by Simons (1995), such proscriptive
able beforehand. It is also irrelevant when asset control may be labelled Boundary Control. And
speci®city is low: in that condition, the market since the feasibility of control instruments is
mechanism takes care of performance evaluation aected by residual, irremediable information
and summarizes all relevant information into impactedness, this variable can still be treated as a
easily accessible price signals. factor that drives the con®guration of control
Non-programmable activities carry a certain instruments.
amount of indeterminacy as a result of uncer-
tainty. This condition may dissolve over time 4.6. Hierarchical Exploratory Control
when in the process of contract execution, infor-
mation accrues on the actual state of the world Exploratory control can be either hierarchical or
and more intimate knowledge on the particulars of hybrid. Which of these alternatives holds, depends
the activities becomes available, allowing the on the degree of asset speci®city. Hierarchical
organization to `recognize the quality of performance Exploratory Control is associated with low pro-
when it sees it'. If these emerging insights spread grammability, high idiosyncrasy and remediable
through the organization, gradually becoming com- information impactedness. In fact, the elements of
mon knowledge, post hoc performance appraisal this archetype can largely be understood as
may be fairly uncontroversial. In this case, the mechanisms that individually and collectively help to
organization is able to evaluate performance using overcome the initial condition of indeterminacy dur-
emergent standards that are shared (or at least ing the process of operationalizing the contribution.
known) by those involved in the process. This is a But before examining these elements, it may be
situation of relatively low ex post information useful to take a closer look at the problematic
impactedness, in which control is expectedly nature of control in the described conditions.
engaged in creating and preserving information At the outset, only vague notions as to the
sharing and in (re)adjusting and (re)aligning per- attainable outcomes of the activity exist. Also,
ceptions on progress. Control structures that can there is insucient knowledge to select in advance
be so described will be referred to as Exploratory the courses of action that are most likely to con-
Control. tribute to satisfactory outcomes, and then to stick
It is important to note that the condition of low to these courses. Since decisions have to be made
post hoc information impactedness is not necessarily anyway, they will be based on the little informa-
an exogenous variable that can be treated as a tion and knowledge that is in fact available. These
given characteristic of the activities in explaining decisions are highly provisional in nature, giving
manifestations of control. It may also be Ð and only broad guidance, and the process of execution
often is Ð the product of control structure choice. of the activity remains an essentially uncharted
There are, however, situations in which ex post route that must be taken one step at the time. This
information impactedness remains incurably high, process is very much a learning process in which
meaning that although the contributor may acquire knowledge and information accrue as to the more
private information on his or her performance, that concrete properties of required contributions, and
information cannot be reliably communicated to on the way in which these can be realized. The
and evaluated by other members of the organiza- accumulating information, however, is likely to be
tion, due to either its specialized character, or to asymmetrically distributed. Moreover, it is likely
432 R.F. Spekle / Accounting, Organizations and Society 26 (2001) 419±441
frequent but ad hoc communication. Yet, as will Performance assessment adopts a long-term
be explained, these same vagaries are precisely the orientation, re¯ecting the period of time required
mechanisms that serve to activate goal-directed for the full eects of decisions and actions to sur-
search behaviour and that necessitate the desired face, and immediate ®nancial compensation for
interaction and sharing of information between some individual superior contribution is, if it exists
participants in the process. at all, of secondary importance. The primary
A typical feature of this archetype is its unwilling- reward structure usually takes the form of a promo-
ness to de®ne and limit individual responsibilities. In tion scheme (including periodic salary increases),
part, this is a predictable response to the impossi- emphasizing long-term performance and encoura-
bility to de®ne in advance what to expect from ging consistently valuable contributions, rather than
those involved in the organization. But it is also a instant but possibly ephemeral success.
means to encourage a problem-solving mentality. Marginson's (1999) study of control practices at
By ``shedding of responsibility as a limited ®eld of Telco Ð a subsidiary of a major British-based
rights, obligations and methods'' (Burns & communications organization operating within a
Stalker, 1961, p. 121), problems are less easily fast-moving, dynamic business environment Ð
referred to other regions of the organization as seems to illustrate some of the points made here
being someone else's responsibility. In the expec- quite tellingly and will therefore be reproduced here
tation that one's eorts will be recognized and at some length. During Marginson's investigation,
valued by the organization, a more rewarding Telco was engaged in a strategic reorientation and
strategy may be to try and solve that problem one- invested rather heavily in innovative projects, giving
self. Frequently, however, solving such problems its activities a signi®cantly non-programmable ¯a-
extends beyond the capacity of the individual, and vour. Learning was a key factor, and the various
he or she must solicit help from other members of projects and programmes were of an acknowl-
the organization. A dependency relation thus builds edged exploratory nature, being subjected to
up. This is in fact a more general characteristic of modi®cation and adjustment as events unfolded.
the hierarchical exploratory control type: indivi- In this setting, Marginson found that although
duals in this archetype are often reliant upon one senior management remained involved in the
another for the accomplishment of their own strategy formation process by sketching the broad
assignments. These mutual interdependencies contours within which initiatives were to ®t and by
reinforce information sharing, cooperation, and emphasizing high priority issues and objectives,
continuous adaptation to new insights as they strategy formulation was to a large extent dele-
emerge, but they also create an incentive to strive gated to lower levels in the managerial hierarchy.
for at least satisfactory performance. Since sub- Strategy formation, thus, very much became a
standard achievement by some individual is not collective exercise and joint responsibility that
exclusively the problem of the organization at involved frequent interaction between managers of
large, but interferes with the performance of direct all levels in the organization, fostering a shared
colleagues on whom the individual depends him- sense of direction and a mutual understanding of
self, opportunistic inclinations (e.g. shirking, were the company was heading. Marginson also
withholding or manipulating information) become reports that instead of relying on formal, vertical
hard to sustain. Moreover, top management is lines of authority, Telco's organizational structure
also involved quite closely in the entire process in was ¯uid and permeable with managers having dif-
a supportive role (reinforcing strategic intentions, ferent roles, serving dierent units, and participating
giving advice, questioning decisions, asking for in dierent teams and projects simultaneously, thus
explanations et cetera). This involvement is valu- facilitating and promoting information sharing and
able in serving coordination and information supporting convergence of insights as to how to
sharing purposes. But in addition, it ensures that proceed. Furthermore, Marginson found that
information relevant for assessment of individual Telco gained cooperation and eort through the
performance reaches the proper hierarchical levels. way in which it organized its activities, featuring
434 R.F. Spekle / Accounting, Organizations and Society 26 (2001) 419±441
a high level of mutual interdependence of man- why as soon as insight into the properties of
agerial activity, thus creating a social pressure to required contributions settles, elements of machine
cooperate. That pressure also ensured that each control gain importance, ultimately to supplant
individual delivered his or her contribution, miti- the exploratory form. But until then, hierarchical
gating opportunistic shirking or manipulative exploratory control may be comparatively ecient,
behaviour. thus explaining its existence.
The resulting common understanding of how to
move the company forward was in fact the major 4.7. Hybrid Exploratory Control
means to achieve coherence in the organization
and to motivate individuals to contribute to the There is, however, a potential alternative. The
emerging agenda. It supplanted most formal Hybrid form of Exploratory Control may oer
mechanisms that commonly serve this purpose. partial relief for the problem of indulgence inher-
Strategy, as described, was a collective endeavour ent in its hierarchical counterpart. Hybrid
rather than an instrument to set preconceived exploratory control is akin to the hierarchical
direction. Although senior management used a variant in its reliance on close interaction, joint
number of key performance indicators, these pro- responsibility, and the resultant information ¯ows
vided but general guidance and were more an to achieve cooperation and behaviour congruence,
attention-directing tool than hard targets to but it provides additional, higher-powered market
achieve, especially at the level of the individual. incentives. These incentives serve to elicit com-
Budgets were not used as motivational devices, mensurate rather than perfunctory performance.
but were just a way to allocate funds. They played More speci®cally, hybrid exploratory control
no role at all in assessment of individual perfor- involves the establishment of outsourcing relations
mance. Other formal instruments of control were with a limited number of suppliers. Due to the
mostly notable for their absence. For instance, high level of uncertainty, contracts must be of a
whereas Telco did have a formal objective-setting general thrust nature and require subsequent
and appraisal system that meant to specify expected operationalization; a process that requires joint
individual contributions and was designed to eorts from supplier and buyer and in which
evaluate achievement in relation to these agreed knowledge and information builds up as to the
upon contributions, Marginson found that this sys- quality that can reasonably be expected. However,
tem was not actually used. In fact, most managers because the buying ®rm engages in more than one
held a rather dismissive attitude towards that system, of these processes simultaneously, a comparative
considering it redundant and irrelevant. assessment of performance becomes possible.
All this seemed to work quite well in the Telco Since it is in the buying ®rm's interest to favour
case. But in general, it must be noted that the the best-performing suppliers when new bids are
hierarchical exploratory control type is also a invited, each supplier has an incentive to oer
markedly indulgent structure, featuring only low- value beyond that of their competitors to acquire
powered incentives. These incentives may be su- a larger share of future business. This structure,
cient to prevent excessively destructive behaviour; which is exempli®ed by the well-known Japanese
they can not be trusted to induce but top-notch outsourcing practices with their merit ranking of
performance. Its reliance on cooperation and suppliers, implicit but credible contract renewal
mutual adjustment foster close personal relations, promises, and strong emphasis on information
which may easily create a lenient atmosphere in sharing between incumbent suppliers (see for
which it is hard to blow the whistle, allowing con- instance Aoki, 1988; Asanuma, 1989; Cooper &
siderable leeway to engage in playful or otherwise Yoshikawa, 1994; Gietzmann, 1996; Hagen &
disfunctional behaviour. Its information ¯ows are Choe, 1998) may, therefore, be more ecient than the
rich but subjective and imperfect. In addition, this hierarchical variant. Viability of hybrid exploratory
archetype's demand for extensive communication control, however, is limited to transactions of
and consultation is resource-consuming. That is moderate speci®city. The higher the idiosyncrasy
R.F. Spekle / Accounting, Organizations and Society 26 (2001) 419±441 435
involved, the less likely it is that suppliers are pre- the actual outcomes some contrast, even for the
pared to invest when the buyer is unable to oer treasurers themselves. And although the depart-
exclusive, long-term contracts. Since exclusive ments were held to report on their activities on a
contracts are obviously unacceptable Ð they regular basis, there was a general feeling that the
would imply full dependency without an enforce- recipients of these reports lacked the level of
able, suciently complete contract and without ®nancial literary required to understand them. In
access to the intricate mechanisms of the hierarchy this setting, Helliar found a tendency to emphasize
to in¯uence behaviour Ð the hybrid form is simply formal rules such as authorization levels and pro-
not on in conditions of high idiosyncrasy. But cedures and policy restrictions (for instance: `we
even when asset speci®city is moderate, there is an don't do naked options'), i.e. a focus on preventing
additional proviso. Exploratory control depends unwanted behaviour3.
signi®cantly on the sharing of private information Boundary control of the hierarchical kind is an
and know-how. Therefore, parties to such an archetype with a distinctly haphazard quality, for
exchange are exposed to the risk of information spill- the information impactedness that de®es perfor-
over. If contracting requires sharing of genuinely mance assessment will also defy a reasonably
important strategic information, it may well be complete ex ante speci®cation of actions to be
that hybrid exploratory control gives way to the avoided. The proscriptions, therefore, tend to be
hierarchical variant, because despite its less pow- limited to unwanted behaviours that were actually
erful incentives, the latter may be more ecient experienced by the organization in the past (and
for its superior information-protecting ability. that were detected on that occasion), to those that
occurred elsewhere and became front-page news,
4.8. Boundary control and dierential asset and to those that just happened to cross top man-
speci®city agement's mind. A related problem arises in
enforcement of the proscriptions. Given the low
Boundary control relates to the governance of observability of actions, it is far from obvious that
non-programmable activities that feature incorri- management is able systematically to detect rule-
gibly high levels of post hoc information impact- breaking behaviour4. Additional safeguards may
edness. This condition implies a seriously limited thus be needed. These can take various forms,
possibility to de®ne and evaluate performance, including periodic external audits and tie-in
even after the contribution has been made. In that arrangements featuring severe sanctions should
situation, there is really not much more one can employees be discovered to cross the line. An
do than to attempt to specify behaviours that are example of the latter would be the practice com-
to be avoided at all cost and to stipulate the monly found in the case of employees charged
boundaries. This is Hierarchical Boundary Control, with some `embezzlement-prone' task, e.g. cashiers,
which is very much the structure of last resort. money and security dealers, and some warehouse-
An interesting example of this archetype can be
found by Helliar in her study of control of the 3
The label boundary control is only appropriate for control
treasury function in 11 companies in the UK of the treasury function as part of the organization at large.
(Helliar, 1998). She found that generally, top Within the treasury department, control was more exploratory
in nature, featuring consultation, joint responsibilities, and
management (and the rest of the organization as
shared expectations.
well) was unable to understand the sometimes 4
In her treasury control study discussed earlier, Helliar
exotic particulars of the treasury process and pro- (1998) for instance found that the Barings collapse had a sig-
ducts. As a result, the treasury departments were ni®cant impact on the tightness of control experienced by the
largely left to themselves. Most companies used treasury departments in her sample, illustrating the `incident-
driven' nature of boundary control. Also, she reports that
some broad performance targets based on a few
typically, de®nition of some of the more important boundaries
pre-de®ned benchmarks, but these had very lim- was largely done by the treasury department itself. Her
ited meaning. Interpreting performance was highly respondents were unanimous in their opinion that it would be
problematic because of lack of standards to give relatively easy to bypass the controls without being found out.
436 R.F. Spekle / Accounting, Organizations and Society 26 (2001) 419±441
personnel. Organizations tend to take an especially specialized legal assistance would seem to qualify
unforgiving attitude towards such employees if as examples. In buying these services, the buyer
they are found to misuse their positions; the most frequently has insucient knowledge to appraise
minor of mistakes being sucient reason for the quality of the services rendered. Yet, the buyer
instant dismissal. With this stern policy, the orga- can Ð to some extent at least Ð rely on reputa-
nization makes sure to reinforce the boundaries, tion eects to assure that the services meet some
signalling that it is quite serious about them, and minimum standard of professionalism. In such
reminding remaining personnel of its commitment markets, reputation is essential to attract clients,
to these rules. But in addition, the employees fre- and not meeting these professional requirements
quently receive a premium wage, i.e. a wage that would expose the supplier to the risk of severe loss
exceeds the amount that would be required to of reputation, and perhaps even a ban from the
mobilize their contributions; this premium serving profession. Although the supplier might be pretty
to increase the employee's stake in respecting the con®dent that sub-standard performance will go
boundaries. undetected in some individual case, he can never
Sometimes, however, de®nition and enforce- be really sure about that. And since the potential
ment of the boundaries can be relegated to the consequences of detection are enormous, the best
market. This requires moderate asset speci®city and policy is simply not to engage in malpractice at all.
the availability of suciently powerful reputation Then, relation-speci®c safeguards and explication
eects in the suppliers market. Audit services and of the boundaries that are to be observed are
Table 1
Control archetypes and their determinants
redundant, and Market-Based Boundary Control is essence of this theory can now be stated more
an ecient archetype. succinctly. The main argument is that Ð given
opportunism and bounded rationality Ð the
4.9. A summary statement speci®c nature of the required contribution Ð to
be expressed in terms of programmability, asset
Having developed and described the contours, speci®city, and ex post information impacted-
concepts, and operative mechanisms of what ness Ð gives rise to distinctive and predictable
amounts to a transaction cost theory of MC, the contractual problems that need to be solved by the
Table 2
Characteristics of the control archetypes
organization. These same attributes also aect the to re®nement, and supports a reasonably detailed
viability of control devices available to cope with study of control issues at the level of organiza-
these problems. The degree of programmability tional subsystems Ð which level certainly is
in¯uences the availability of norms and standards to appropriate for research in MC. At that analytical
direct behaviour and, consequently, the feasibility level, this paper has proposed a transaction cost
and strength of prescriptive control. Asset speci®city theory of MC. This theory intends to explicate the
is relevant for it de®nes access to market-based link between various archetypical con®gurations
incentives and, more generally, the mechanisms of control devices and the kind of activities they
available to cope with opportunism. The degree of are expected to control. The nature of the organi-
ex post information impactedness denotes the zational activities and the contributions from
possibility (or impossibility) to transform in the organizational participants that are required to
process of execution of the activity a condition of ex perform these activities can be discriminatingly
ante uncertainty into a situation of shared under- de®ned through their scores on three dimensions:
standing of what constitutes good performance and (1) the extent to which the contributions are sus-
how such performance can be delivered. These ceptible to up front programming; (2) the degree
variables and their eects on the ecacy of indi- of asset speci®city; and (3) the intensity of ex post
vidual instruments of control are used to identify information impactedness. Given bounded ration-
nine distinct control archetypes or consistent clusters ality and opportunism, these features are predictably
of control devices. The control archetypes dier in associated with distinctive control problems that
their problem-solving ability and in respect of cost, need to be dealt with. The control archetypes Ð
and the alignment between a control archetype and a nine of which have been identi®ed and described Ð
required contribution is explained by demonstrating dier in their problem-solving ability, which
the archetype's comparatively ecient ability to makes them appropriate for the governance of
deal with the contractual problems inherent in the some contributions, but not for others. Moreover,
contribution which it is supposed to control. they dier in respect of cost, and ultimately, an
Tables 1 and 2 summarize the details. empirically observable alignment of a contribution
with a control archetype is held to be explainable
by delineating the relative eciency properties of
5. Summary and discussion the match, either quantitatively or Ð more
likely Ð in a qualitative way.
This paper has argued that the discipline of MC It would seem that this theoretical approach has
is likely to bene®t from theories that provide a a good deal to recommend it. If MC is about
cogent and comprehensive perspective to address enhancing organizational eectiveness Ð and that is
the issue of MC structure variety, and examined a generally accepted position Ð , MC-theory should
the potential of TCE to inform such theories. TCE specify the modus operandi of MC-structures in
approaches phenomena of organization from a delivering their contribution, and it must be able
comparative point of view in which dierent to evaluate these structures explicitly in terms of
organizational arrangements are studied as alter- (dierential) functionality. Especially this last
native ways to govern economic transactions. aspect has received remarkably little attention in
Essentially, TCE is about understanding the MC-theorizing, and eectiveness is more usually
match between activities to be controlled and assumed than demonstrated. TCE's remediableness
control structures, and holds that the character- criterion moves beyond paying lip-service to eec-
istics of the activities and the credentials of the tiveness, and oers a reasonably concrete and prac-
structures determine and explain which structures ticable procedure to address this issue. Furthermore,
are appropriate and which are not. TCE shares its the proposed theory suggests a pragmatic way to
central problem Ð explaining control Ð with handle the issue of de®ning the organizational
MC, albeit that the latter requires a higher level of goals that MC is supposed to serve. These goals
resolution. The logic of TCE, however, is receptive are notoriously elusive, but they can be presumed
R.F. Spekle / Accounting, Organizations and Society 26 (2001) 419±441 439
to aect the more easily identi®able patterns of variables is markedly rough and the boundaries of
actions, decisions and beliefs that shape what the the intervals are left implicit. And the archetypes
organization is doing. The presented approach are constructs that (presumably) help to recognize
concentrates on the actual activities in which the and expound general tendencies, but they are not
organization is engaged, but studies these in rela- fully compelling categories. When applied in an
tion to the patterns of which they are part to empirical setting, the archetypes may not be descrip-
include in the analysis an understanding of the tively accurate in every respect, and the observed
meaning the organization attaches to these activ- con®guration of control may not fall neatly into any
ities. Such an understanding provides the means to of the pre-identi®ed classes. For these reasons, the
relate concrete activities and contributions to what application of this theory is bound to command con-
the organization apparently is trying to achieve, siderable interpretative eorts from the researcher to
and allows to dierentiate realized from desired deal with the inevitable shades of grey. However, I
activities and contributions. The proposed frame- submit that the approach is ¯exible enough to cope
work, then, focuses on the way in which MC with the ambiguities, and that it is suciently cogent
helps to reduce the gap that may arise between the to ensure satisfactorily restrictive interpretations
two in a cost-eective manner. This is a shame- that preserve the general logic of the argument
lessly instrumental approach in that it ignores and on which academic consensus can be reached.
more high-brow questions as to the contribution Anyway, since TCE itself suers from similar vag-
of MC to overall organizational eectiveness. But aries, and since these have by no means impeded
it is operational, and this mere fact puts the empirical application and testing (see Shelanski &
approach ahead of most alternative approaches Ð Klein (1995) and Masten (1996) for recent over-
at least of those with similar wide-ranging views of empirical research in TCE) there seems to
ambitions. be no basis to be particularly wary, and the best
The reference to these ambitions suggests way to go is probably just to give it a try.
another asset of the transaction cost approach to
MC. Whereas traditionally, the focus of MC has
been on the individual organization, there is a Acknowledgements
growing concern that this focus may be too
restrictive (cf. Berry, 1994; Otley, 1994; Otley, I thank Marcel Bonnet, Jan van Helden, Rien
Broadbent & Berry, 1995). Since coordination and van Hoepen, Chris Knoops, David Otley, and Ed
control of economic activity is increasingly often Vosselman for their encouragement and suggestions.
the province of all kinds of collaborative struc- Also, the participants in the workshop on the
tures between ®rms (e.g. strategic alliances, supply Economics of Management Accounting and Control
networks, and joint ventures), the scope of MC- held in Rotterdam in October 2000 and the anon-
theory must be broadened to allow coverage of ymous reviewers provided valuable comments.
these arrangements. The theory proposed in this
paper Ð and TCE as its intellectual ancestor Ð
are well-positioned to answer to this challenge, for References
they oer an integrative and symmetrical point of
view from which to address issues of control of Ansari, S. L. (1977). An integrated approach to control systems
design. Accounting, Organizations and Society, 2, 101±112.
any kind, irrespective of the particular organiza- Anthony, R. N. (1965). Planning and control systems. Boston:
tional arrangement in which they arise. That is Harvard Business School Press.
undoubtedly a welcome prospect. Anthony, R. N. (1988). The management control function. Bos-
Finally, the proposed theory is empirically ton: Harvard Business School Press.
Aoki, M. (1988). Information, incentives, and bargaining in the
testable Ð in principle at least. To be sure, the
Japanese economy. Cambridge: Cambridge University Press.
variables have been described at a conceptual level Asanuma, B. (1989). Manufacturer-supplier relationships in
and their substance is suggested rather than Japan and the concept of relation-speci®c skill. Journal of the
de®ned. Moreover, the scale on which to score the Japanese and International Economies, 3, 1±30.
440 R.F. Spekle / Accounting, Organizations and Society 26 (2001) 419±441
Baiman, S. (1982). Agency research in managerial accounting. Hagen, J. M., & Choe, S. (1998). Trust in Japanese inter®rm
Journal of Accounting Literature, 1, 154±213. relations: institutional sanctions matter. Academy of Man-
Baiman, S. (1990). Agency research in managerial accounting: agement Review, 23, 589±600.
a second look. Accounting, Organizations and Society, 15, Hartmann, F. G. H. (2000). The appropriateness of RAPM:
341±371. toward the further development of theory. Accounting,
Berry, A. J. (1994). Spanning traditional boundaries: organiza- Organizations and Society, 25, 451±482.
tion and control of embedded operations. Learning and Helliar, C. (1998). Management control of the treasury func-
Organizational Development Journal, 15(7), 4±10. tion. Paper presented at the BAA's 1998 Annual Conference,
Berry, A. J., Broadbent, J., & Otley, D. (1995). Management con- Manchester.
trol: theories, issues and practices. London: Macmillan Press. Hopwood, A. G. (1978). Towards an organizational perspec-
Birnberg, J. G., Turopolec, L., & Young, S. M. (1983). The tive for the study of accounting and information systems.
organizational context of accounting. Accounting, Organiza- Accounting, Organizations and Society, 3, 3±13.
tions and Society, 8, 111±129. Hopwood, A. G. (1983). On trying to study accounting in the
Burns, T., & Stalker, G. M. (1961). The management of inno- contexts in which it operates. Accounting, Organizations and
vation. London: Tavistock Publications. Society, 8, 287±305.
Carruthers, B. G. (1995). Accounting, ambiguity, and the new Johnson, H. T. (1983). The search for gains in markets and
insitutionalism. Accounting, Organizations and Society, 20, ®rms: A review of the historical emergence of management
313±328. accounting systems. Accounting, Organizations and Society,
Chapman, C. S. (1997). Re¯ections on a contingent view of 8, 139±146.
accounting. Accounting, Organizations and Society, 22, 189± Lang®eld-Smith, K. (1997). Management control systems and
205. strategy: a critical review. Accounting, Organizations and
Chua, W. F., Lowe, T., & Puxty, T. (1989). Critical perspectives Society, 22, 207±232.
in management control. Houndmills: The Macmillan Press. Lowe, E. A. (1971). On the idea of a management control sys-
Cooper, D. J., & Hopper, T. M. (1990). Critical accounts. tem: integrating accounting and management control. Jour-
Houndmills: The Macmillan Press. nal of Management Studies, 8, 1±12.
Cooper, R., & Yoshikawa, T. (1994). Inter-organizational cost Lowe, T., & Chua, W. F. (1983). Organisational eectiveness and
management systems: the case of the Tokyo-Yokohama- management control. In T. Lowe, & J. L. J. Machin, New per-
Kamakura supplier chain. International Journal of Produc- spectives in management control (pp. 271±287). New York: St.
tion Economics, 37, 51±62. Martins Press.
Covaleski, M., & Aiken, M. (1986). Accounting and theories of Lowe, T., & Puxty, T. (1989). The problems of a paradigm: a
organizations: some preliminary considerations. Accounting, critique of the prevailing orthodoxy in management control.
Organizations and Society, 11, 297±319. In W. F. Chua, T. Lowe, & T. Puxty, Critical perspectives
Covaleski, M. A., Dirsmith, M. W., & Samuel, S. (1996). in management control (pp. 9±26). London: Macmillan
Managerial accounting research: the contributions of orga- Press.
nizational and sociological theories. Journal of Management Marginson, D. E. W. (1999). Beyond the budgetary control
Accounting Research, 8, 1±35. system: towards a two-tiered process of management control.
Cyert, R. M., & March, J. G. (1963). A behavioral theory of the Management Accounting Research, 10, 203±230.
®rm. Englewood Clis: Prentice Hall. Masten, S. E. (1996). Empirical research in Transaction Cost
Dermer, J. (1988). Control and organizational order. Account- Economics: challenges, progress, directions. In J. Groenewe-
ing, Organizations and Society, 13, 25±36. gen, Transaction cost economics and beyond (pp. 43±64). New
Fisher, J. (1995). Contingency-based research on management York: Kluwer Academic Publishers.
control systems: categorization by level of complexity. Jour- van der Meer-Kooistra, J. (1994). The coordination of internal
nal of Accounting Literature, 14, 24±53. transactions: the functioning of transfer pricing systems in
Fisher, J. G. (1998). Contingency theory, management control the organisational context. Management Accounting Research,
systems and ®rm outcomes: past results and future direc- 5, 123±152.
tions. Behavioral Research in Accounting, 10, 47±64. Merchant, K. A. (1982). The control function of management.
Flamholtz, E. G. (1983). Accounting, budgeting and control Sloan Management Review, Summer, 43±55.
systems in their organizational context: theoretical and Merchant, K. A. (1985). Control in business organizations. New
empirical perspectives. Accounting, Organizations and York: Ballinger Publishing Company.
Society, 8, 153±169. Merchant, K. A., & Simons, R. (1986). Research and control in
Flamholtz, E. G. (1996). Eective management control: theory complex organizations: an overview. Journal of Accounting
and practice. Boston: Kluwer Academic Publishers. Literature, 5, 183±203.
Galbraith, J. (1973). Designing complex organizations. Reading: Mintzberg, H. (1983). Structures in ®ves: Designing eective
Addison-Wesley Publishing Company. organizations. Englewood Clis: Prentice Hall.
Gietzmann, M. B. (1996). Incomplete contracts and the make Mintzberg, H., & Waters, J. A. (1985). Of strategies, deliber-
or buy decision: governance design and attainable ¯exibility. ate and emergent. Strategic Management Journal, 6, 257±
Accounting, Organizations and Society, 21, 611±626. 272.
R.F. Spekle / Accounting, Organizations and Society 26 (2001) 419±441 441
Mouritsen, J. (1999). The ¯exible ®rm: strategies for a sub- Simons, R. (1990). The role of management control systems in
contractor's management control. Accounting, Organizations creating competitive advantage: new perspectives. Account-
and Society, 24, 31±55. ing, Organizations and Society, 15, 127±143.
Otley, D. T. (1980). The contingency theory of management Simons, R. (1995). Levers of control: How managers use inno-
accounting: achievement and prognosis. Accounting, Organi- vative control systems to drive strategic renewal. Boston:
zations and Society, 5, 413±428. Harvard Business School Press.
Otley, D. T. (1983). Concepts of control: the contribution of Spicer, B. H., & Ballew, V. (1983). Management accounting
cybernetics and systems theory to management control. In T. systems and the economics of internal organization.
Lowe, & J. L. J. Machin, New perspectives in management Accounting, Organizations and Society, 8, 73±96.
control (pp. 59±87). New York: St. Martins Press. Swieringa, R. J., & Waterhouse, J. H. (1982). Organizational
Otley, D. T. (1984). Management accounting and organization views of transfer pricing. Accounting, Organizations and
theory: A review of their interrelationship. In R. W. Scapens, Society, 7, 149±165.
D. T. Otley, & R. J. Lister, Management accounting, organi- Tiessen, P., & Waterhouse, J. H. (1983). Towards a descriptive
zational theory and capital budgeting: Three surveys (pp. 96± theory of management accounting. Accounting, Organiza-
164). London: Macmillan Press. tions and Society, 8, 251±267.
Otley, D. T. (1994). Management control in contemporary Vaivio, J. (1999a). Examining ``the quanti®ed customer''.
organizations: towards a wider framework. Management Accounting, Organizations and Society, 24, 689±715.
Accounting Research, 5, 289±299. Vaivio, J. (1999b). Exploring a `non-®nancial' management
Otley, D. (1999). Performance management: a framework for accounting change. Management Accounting Research, 10,
management control systems research. Management 409±437.
Accounting Research, 10, 363±382. van der Meer-Kooistra, J., & Vosselman, E. G. J. (2000).
Otley, D., Broadbent, J., & Berry, A. (1995). Research in man- Management control of inter®rm transactional relationships:
agement control: an overview of its development. British the case of industrial renovation and maintenance. Account-
Journal of Management, 6, S31±S44. ing, Organizations and Society, 25, 51±77.
Ouchi, W. G. (1977). The relationship between organizational Williamson, O. E. (1975). Markets and hierarchies: Analysis and
structure and organizational control. Administrative Science antitrust implications. New York: Free Press.
Quarterly, 22, 95±113. Williamson, O. E. (1979). Transaction Cost Economics: the
Ouchi, W. G. (1979). A conceptual framework for the design of governance of contractual relations. Journal of Law and
organizational control mechanisms. Management Science, Economics, 22, 233±261.
25, 833±848. Williamson, O. E. (1985). The economic institutions of capital-
Ouchi, W. G. (1980). Markets, bureaucracies, and clans. ism. New York: Free Press.
Administrative Science Quarterly, 25, 129±141. Williamson, O. E. (1993). Opportunism and its critics. Man-
Rotch, W. (1993). Management control systems: one view of agerial and Decision Economics, 14, 97±107.
components and their interdependence. British Journal of Williamson, O. E. (1996). The mechanisms of governance. New
Management, 4, 191±203. York: Oxford University Press.
Scapens, R. W. (1991). Management accounting: A review of Williamson, O. E. (1999a). Public and private bureaucracies: a
contemporary developments (2nd ed.). London: Macmillan Transaction Cost Economics perspective. Journal of Law,
Education. Economics and Organization, 15, 306±342.
Seal, W. B. (1993). Accounting, management control and busi- Williamson, O. E. (19999b). Strategy research: governance and
ness organisation. Aldershot: Avebury. competence perspectives. Strategic Management Journal, 20,
Shelanski, H. A., & Klein, P. G. (1995). Empirical research 1087±1108.
in transaction cost economics: a review and assessment. Williamson, O. E., & Ouchi, W. G. (1981). The markets and
Journal of Law, Economics, and Organization, 11, 335± hierarchies program of research: origins, implications, pro-
361. spects. In A. H. Van de Ven, & W. F. Joyce, Perspectives on
Simon, H. A. (1945). Administrative behavior (3rd ed. 1976). organization design and behavior (pp. 347±370). New York:
New York: Free Press. John Wiley & Sons.