Interdependencies, trust and information in relationships,alliances and networks

Cyril Tomkins *

School of Management, University of Bath, Bath BA2 7AY, UK

Abstract

The last decade has been characterised by the increased development of cross-company relationships, alliances and

more complex business networks. This paper examines fundamental concepts that relate to the needs for information,including accounting information, in these interactive structures. It considers, initially, some consequences foraccounting when planning and control is to be exercised across organisational boundaries, but the main thrust of thepaper is to focus on the fact that all relationships depend to some extent on trust. The interaction between trust and

information in personal relationships is explored to serve as a template against which to consider whether the infor-mation needs of inter-organisational relationships are similar. Having examined information needs of business rela-tionships in general, the analysis is developed to consider the needs of di�erent forms of corporate alliances and then

wider business networks. This paper is an attempt to create a debate on new questions that seem to be fundamental tothe design of information and accounting systems. It is not claimed to have provided de®nitive answers, but it is pro-posed that the lack of a well-developed theory of the interaction between trust and information has been a fundamental

gap in our knowledge. # 2000 Elsevier Science Ltd. All rights reserved.

This paper is the result of setting out to considerhow the growth of business alliances and networksimpact upon business information, includingaccounting. Recent discussion of business alliancesand networks implies a much closer relationshipbetween di�erent parties than is involved in occa-sional buying and selling. It recognises that even aregular pattern of trading creates social bondsbetween parties based upon mutual understandingand trust and, consequently, lower transactionscosts.Moreover, the interactions between companiesoften develop much further into supply chain

``partnerships'' with an implied sense of sharing inknowledge, decision-making and the collectiverewards. Furthermore, these alliances often existwithin a wider network of business involvementthat possesses within it a range of relationships ofdi�erent type and intensity. It is recognised thatwhat one means by the words `alliance' or `net-work' is not unproblematic. There is a considerablevariation in types of alliances (as described later inthis paper) as well as di�erent emphases in under-standing what business networks are. Debressonand Amesse (1991) and Tidd, Bessant and Pavitt(1997) argue, for example, that networks should notbe de®ned simply as transaction `webs' or `clusters'of organisations and Castells (1996, pp. 157±163)discusses di�erent forms of networks. Also Tidd et

0361-3682/00/$ - see front matter # 2000 Elsevier Science Ltd. All rights reserved.

PI I : S0361-3682(00 )00018-0

Accounting, Organizations and Society 26 (2001) 161±191

www.elsevier.com/locate/aos

* Corresponding author. Tel.: +44-1225-826683; fax; +44-

1225-826473.

E-mail address: c.r.tomkins@bath.ac.uk

al. use the term networking to apply speci®cally tothe development of long-term innovation networkswhich they clearly distinguish from the variousother forms of alliances. These insights are valuablefor warning that one should not expect the infor-mation needs for all alliances or all business net-works to be the same. It was decided, however, toleave this issue until later in the paper. It seemedmore useful ®rst to:

(1) undertake a brief consideration of somebasic points relating to how accounting andbusiness information is used when trying tomaster events1 across more than one organi-sation and

(2) provide a more extensive consideration ofhow the provision of information needs totake into account the strength of socialbonds or trust that exist in an alliance.

These more fundamental issues need to beexamined ®rst to provide a ®rmer basis for con-sidering di�erences in information requirementsbetween types of alliances and networks.

1. Mastering events across organizational bound-aries: some basic points

It is quite obvious, without any initial considera-tion of social bonds or trust, that accounting ana-lyses for alliances have to capture e�ects through atleast two organisations, rather than one. For exam-ple, where two or more ®rms manufacture productscollaboratively, problems in understanding pro-duct costs might arise over inconsistent bases ofoverhead allocation. In fact overheads in oneorganisation might drive overhead expenditure inanother such that activity based or feature costingexercises could become more complex. In princi-ple, cost behaviour analyses and target costingexercises will also have to take these cross-®rme�ects into account. Nicolini, Tomkins, Holti,

Oldman and Smalley (2000) illustrate the problemsthat can arise in attempting to develop targetcosting across the supply chain in the constructionindustry, while Barber, Graves, Hall, Sheath andTomkins (2000) refer to similar cross-organisa-tional problems that can arise in undertaking costof quality exercises.Responsibilities and budgets might also run

across organisational boundaries and capacitylimits in one ®rm could a�ect the other, giving riseto the need to construct the inter-organisationalbudget before the organisation's own budgets canbe constructed. To facilitate collaborative budget-ing on alliances, it might also help to move to anincrease in the use of projects as a basis for codingexpenditure and revenue in the regular budgetingand accounting processes as well as in investmentappraisal.Cost reduction activities might also switch

emphasis over time between organisations. Acomprehensive kaizen approach to continuingimprovement might simply mean that all in bothorganisations continually seek marginal improve-ments. In contrast, preference for a theory ofconstraints approach to e�ciency improvement(Goldratt, 1990; Tomkins, 1999) which focuses onrelaxing the dominant constraint at any point oftime might see that focus switching back and forthacross the partners to the alliance bringing, per-haps, extra management issues to be addressed.Financing issues such as ®nancial gearing and

covenants on loans might also take on a multi-organisational dimension, as would also investmentappraisal. If two ®rms are dependent on each otherto complete a project, both ®rms will be interestedthat the other does not adopt too risky a positionwith regard to all its activities. The risk attached tothe other party with whom one is deeply involvedcould impact upon one's own ®nancial risk asperceived by the market and, hence the cost ofcapital and ability to sustain debt levels. The col-lapse of the other ®rm might also entail projectfailure or at least signi®cant reorganisation costsand delays in ®nding new partners.In considering these fairly obvious consequences

of multi-organisation collaboration, there seemsto be no obvious need to develop new accountingtechniques. Rather there is the need for cross-

1 The term `mastery of events' is used deliberately here to be

consistent with Luhmann's (1979) use of the term. The sig-

ni®cance of this will become clear later in the paper.

162 C. Tomkins / Accounting, Organizations and Society 26 (2001) 161±191

organizational design of management accountingsystems. It is doubted, however, if many compa-nies yet go to the extent suggested here in cross-organisation cost management, budgeting and®nancial leverage/cost of capital analyses and, nodoubt, complexities attached to tracing suche�ects across two or more organisations willemerge as more research is conducted. This is arich ®eld for empirical research. It is clear already,however, that, with a movement to collaborationacross organisations, there must be a greateremphasis on negotiation to determine terms ofthat co-operation. To some extent negotiation ispresent between all organisations trading witheach other and between levels within an organisa-tion, but the development of long-standing, deepalliances implies sharing information and workingout collaborative futures showing how the risk/return position of each is a�ected by di�erentactions. It is likely that the accountants in all mem-ber companies of an alliance will need to becomemore closely involved with each other in the nego-tiation of actions and pro®t-sharing. Moreover,negotiation is normally an iterative procedurethrough which a deeper understanding of the nat-ure of the project and one's part in it come to beunderstood. Large projects are not just ``there''with given costs and bene®ts to be shared. Theyhave to be shaped and developed with modi®ca-tion of partner's roles and interests. This processwill be facilitated by repeated simulation of eachparty's risk/return position until agreement isreached. In such a process, the accounting infor-mation needed is itself a re¯exive product of thedeveloping negotiation. A growing emphasis onsimulation especially linked to investment deci-sions and new product development (target cost-ing) might therefore be expected.Discussion of pro®t sharing in existing account-

ing literature is largely focused upon the internaltransfer pricing issue. Once one recognises theexistence of close on-going relationships and jointplanning between separate businesses, it is likelythat the traditional recipes for transfer pricing willbe quite inadequate as the basis for pro®t sharingbetween the parties involved. While collaborativerisk analysis and fair pro®t sharing schemes arementioned in the literature (e.g. Monden, 1989;

Tomkins, 1991;2 Tomkins & McAulay, 1992),transfer-pricing literature traditionally lookedupon the calculation of optimal marginal cost-based transfer prices and negotiation as alternativemeans of solving the transfer-pricing problem.Such a view ignores the process by which negotia-tion takes place in situations where the terms ofcollaboration agreed are likely to have a sig-ni®cant impact upon the participating parties.Increasingly, in business alliances, this involvescollaboration over development and investmentand not just trading on the basis of existing goods.Negotiation then needs to be supported by revisedrequests for information and revised calculationsof costs and bene®ts to each party. The implica-tion of this is that it will become necessary to takethe collaborative decision based on two levels ofanalysis. First, the aggregate investment itselfmust earn the required rate of return for the riskattached to it, irrespective of how many parties areinvolved. Second, each party will not participatein the project unless it sees the prospect of fair3

rewards for the risks it faces for its individualcontribution to the project.To continue with a detailed consideration of

such simulations would take us too far away fromthe main thrust of argument in this paper, but a verysimple example of simulation in the Appendixattempts to illustrate the general nature of this inter-action between revised information, revised calcula-tion and negotiation. The example demonstrateshow each party gradually develops understandingof what involvement in the project entails and theway in which a mutually satisfactory basis forprogress emerges. This is how the presence of

2 Tomkins (1991, Chapter 9) argued that organisations con-

stitute a negotiated environment and that managing a ®rm

involves gaining the support of various di�erent factions that

could be de®ned as ``relevant entities'' and that collaboration

would not be gained without assessing the risks and likely

returns faced by each of the main entities.3 The terms ``required rate of return'' and ``fair rewards''

are, of course, problematic. To avoid a long digression here, it

will be assumed that collaborating parties will have their own

perceptions of what is required. Although they may take some

guidance from conventional ®nance literature, it is expected

that further research might well demonstrate that these per-

ceptions themselves are usually the output of a negotiation

process and not easily formulated in advance.

C. Tomkins / Accounting, Organizations and Society 26 (2001) 161±191 163

deeper inter-party relationships is di�erent fromsimple market buying and selling. Information isnot given on a take it or leave it basis. Relevantinformation is itself a function of the interactiveprocess. As a collaborative venture is planned, itwill be necessary to provide new information andrepeated detailed analysis. This will be requiredespecially at the stage in the alliance where majorjoint investment activities are being considered4

and, in practice, the interaction is likely to besomewhat more complex than it is in this example.Basic accounting techniques may not need revi-sion, but the way in which accountants perceivetheir roles and employ their techniques and infor-mation bases may well change. The area warrantsmore empirical research with a greater emphasisupon business processes and the use of accountingin action/negotiation.

2. The relationship between trust and information

While there is some merit in highlighting thisincreased emphasis on negotiation and interactionin business alliances, and the associated practicalproblems of reliably simulating pro®t sharing invery complex projects, more fundamental issuesarise when the nature of the social bonds and trustare explored and related back to the need forinformation. This will be the main focus of the restof the paper.

2.1. Hierarchical levels of collaboration

A fairly simple structure is adopted for the restof this article that di�erentiates between three dif-ferent hierarchical levels of collaboration: rela-tionships, alliances and networks. A relationship istaken as the bedrock upon which any alliance isformed, but one needs to distinguish relationshipsfrom alliances because a range of di�erent forms

of corporate alliances exist which may have dif-ferent implications for business information. Anetwork implies, however, rather more than a justa bilateral relationship or even an alliance. Net-works are much more complex and formed fromcon®gurations of alliances and relationships thatrange from intimate partnerships to simply buyingand selling on a competitive basis or even justexchanging views and other information. As Tiddet al. say:

An organisational network can be thought ofconsisting of a number of positions or nodes,occupied by ®rms, business units, universities,governments, customers or other actors, andlinks or interactions between these nodes.(Tidd et al., 1997, p. 210).

and, following Castells (1996), one can also thinkof large corporations as networks in themselves,where they have replaced their traditional hier-archical organisation structures with a highlydecentralised form with many autonomous units,and di�erent parts of a large organisation will belinked into di�erent external networks.This paper will develop propositions relating to

business information at all three levels just speci-®ed. It will commence with a consideration ofsome general theorising relating to informationand control needs in relationships. This will thenbe applied to specifying likely information needsin evolving personal relationships. It was felt thata consideration of the use of information at thislevel of social interaction would provide a clar-i®cation of basic concepts. The paper will then con-sider to what extent the determinants of informationneeds at a personal level are applicable to relation-ships between organisations. It will then expand thediscussion to consider the distinct information needsof di�erent types of alliances. Finally, the analysiswill be raised to consider what is needed at thenetwork level. Close relationships at a personallevel depend to a signi®cant extent on the estab-lishment of trust. It is, therefore, important toaddress, ®rst, the concept of trust to see why it isso central to the question of information proces-sing within relationships.

4 As a by-product, the example in the Appendix also illus-

trates how large errors in pro®t and risk sharing can occur if

accountants simply rely on discounting expected values (or

®xed-point estimates of cash ¯ows). This may well lead to later

disappointment and even suspicion about honesty in negotia-

tion and collapse of the project.

164 C. Tomkins / Accounting, Organizations and Society 26 (2001) 161±191

2.2. De®ning trust and its centrality for de®ninginformation need

Luhmann (1979) argues that trust in its broadestsense is simply having con®dence that one's expec-tations will be realised. At that level of analysis, trustcan apply to things as well as people and their sys-tems. Luhmann states that, without trust, we couldhardly get up in the morning.5 At root, trust is sim-ply an assumption of functional continuity of thatwith which we are familiar. This is really no di�er-ent from any schema that we all adopt about howthe world works that can be bracketed o� to let usfocus upon what we cannot take as given and needsdealing with right now [see Gioia (1986), for exam-ple, on the centrality of schemas to everyday life].The discussion in this paper concerns trust that

we have in people and systems run by people andyet, at the most fundamental level, this arises fromthe same basic requirement as trust in things.Trust in people also enables us to adopt schemaswhich leave us free to act without trying to processmore information about of the world than we arecapable of doing. Trusting in someone enables usto act as if the uncertainty that we face is reduced,although it does not reduce that actual uncer-tainty. Trust is, therefore, ubiquitous: it is a fun-damental building block of social life. Somedegree of trust in what things are and what can bedone with them is unavoidable, even if we live thelife of a hermit. If one wants to understand howany relationship works, one must, therefore,address the boundaries of trust within that rela-tionship. For this purpose trust will be de®ned as:

The adoption of a belief by one party in arelationship that the other party6 will not act

against his or her interests, where this belief isheld without undue doubt or suspicion and inthe absence of detailed information about theactions of that other party.

The word `undue' is important because we cannever remove all uncertainty about a future actand there cannot be signi®cant trust unless there isalso freedom to break that trust, even if one doesnot expect such a breach. The de®nition alsohighlights the fact that trust implies adopting abelief without full information. The notions oftrust and the design of information requirementsare, therefore, inextricable linked. If we accept thisbroad understanding of trust, namely that de®ninga `schema of givens', the state of trust at any timeis the fundamental determinant of informationrequirements. A person or organisation seeking toexercise mastery over some a�airs will requireinformation on only those matters that it has beendecided not to take on trust. This decision will bethe function of the perceived uncertainty attachedto the functioning of a person or thing plus theconsequences of that non-functioning.The point is so fundamental that it is surprising

that it has not taken a much more explicit anddominant role in our theories about informationsystems, including accounting.7 While, the per-ceived level of uncertainty and possible adverseconsequences in¯uence the balance between trustand information needed for coping with uncer-tainty, it must be more clearly recognised that theexercise of trust is an alternative uncertainty

5 I take this to mean that we have trust that we will be able

to stand when we get out of bed, that we have trust that the tap

will contain water, that the car is still in the garage and will

start and that I will arrive at work safely. A colleague, Thomas

Ritter, pointed out that trust in manufactured things is really

trust in the person who made them performing his/her job

properly and so trust in man-made things and people may not

fundamentally di�erent.6 This de®nition is framed in terms of trust in people.

Obviously inanimate objects have no intent, although, if they

fail to operate as intended, they can cause problems as if the

object acted against the person a�ected.

7 It is recognised that the theme of trust versus control is

evident in debates such as that addressed by Power (1994, 1996,

1997), but the point being made here is that the notion of the

balance between trust and the use of control mechanisms might

be the starting point for the design of all information systems,

not just that related to audit and public accountability. More-

over, Power does not really address change in terms of devel-

oping trust, but rather in changing supervision by involvement

in processes, although, of course, that might lead to greater

trust. Furthermore, the demands over the last decade for more

and more information and more and more accountability both

within and between organisations is tantamount to a decrease

in trust. Recti®cation of unsatisfactory performance, in so far

as it exists, has rarely been expressed in terms of what could be

done to restore trust and yet this might be far more cost e�ec-

tive not only in terms of avoiding information processing costs,

but also in stimulating improved activity by those assessed.

C. Tomkins / Accounting, Organizations and Society 26 (2001) 161±191 165

absorption mechanism to increased information.The role of information and accounting in uncer-tainty reduction is obvious and widely discussedand analysed, especially in information economicsapproaches (see Feltham, 1968; Feltham &Demski, 1969), but the role that developing morereliable forms of trust and the cost of doing that israrely considered in the cost±bene®t analysis.8 Infact, it might not be too extreme to argue that infor-mation and accounting system designers shouldalways commence the design of their systems byconsidering carefully and explicitly what and whocan or cannot be trusted. This point alone seemsto open up a stream of research possibilities rightacross the whole realm of accounting although thisis not the place to undertake such an analysis.9

2.3. A functional relationship between trust andinformation need and the rational basis for trust

Trust should be, then, a pivotal concept whenaddressing uncertainty reduction, the provision ofinformation and accounting theory. Wicks, Ber-man and Jones (1999) recently took up this themewith regard to general business information andargued that businesses should attempt to identifythe optimal level of trust that they employ in theirbusiness relationships. They say that there must bean optimal level of trust. If excessive trust isplaced in the business partner, one may be takingtoo much risk for ones own wishes without fullyunderstanding it. This may lead to resources beingmisallocated from one's own point of view andone might even be acting naively such that thepartner can take advantage. If, on the other hand,there is little trust, but one still wishes to deal withthe other party, there will be the need for morecomplete information about plans (what the other

party intends to do), processes (how it is proposedto do it) and results (desired or unexpected endsobtained) coupled with appropriate sanctions forinappropriate behaviour. This, Wicks et al. argue,will incur higher agency costs10 and may have

8 A notable exception is Creed and Miles (1996), who

recognise that there are costs involved in both developing and

operating control mechanisms and in building and maintaining

trust, but they do not follow through the implications for

information provision through in detail.9 It is interesting to note that Luhmann actually refers to

some of our own accounting literature and refers to Wildavsky

(1964) who showed that Congressional budgets were far too

complex for Congressmen to understand in detail and who,

therefore, needed rather to decide what they could and could

not trust.

10 It has been suggested to the author that this debate about

trust versus information is simply a restatement of the agency

theory debate. This is far from being the case. Agency theory is

concerned with the formulation of appropriate incentives to

motivate a subordinate to operate in the way a superior

requires; it is not concerned with identifying a fair share of the

gains. It o�ers such incentives as an alternative to the provision

of full and accurate information that, due to moral hazard and

information asymmetries on the agent's actions, may not be

attainable in reliable form anyway. Agency theory assumes that

the subordinate will act against the principal's interest unless

incentivised not to do so. Also, the operation of agency theory

incentives depends upon the ability of the superior to specify, in

advance, desired outputs in unambiguous form and strictly

correlate the agent's rewards with that output. Managing rela-

tionships through increased trust does not presume a master

and agent relationship, is concerned with creation of fair shares

and reciprocity and o�ers a di�erent route to overcoming

moral hazard and information asymmetries. It stems from quite

a di�erent philosophy. Moreover, the emerging and interactive

nature of decisions in alliances with the iterative development

of relevant information and insights has been emphasised. This

seems to be precisely the situation where agency theory is of

limited value, especially when alliances have the aim of devel-

oping a relationship from which the longer-term bene®ts can-

not be precisely predicted or assessed. This is not to denigrate

agency theory. It has an important place in control regimes

where there is a clear master and agent relationship, desired

outputs are clear and measurable and incentives can be unam-

biguously correlated with the outputs. Agency theory may well,

therefore, have a role to play in developing control over short-

run relationships where ultimate power to appoint rests mainly

on one side of that relationship. Trust mechanisms are more

appropriate at the other end of the continuum where there is

longer-term collaboration and less clear power dominance and,

moreover, where the bene®ts and costs, and attitudes of the

parties towards them, are less distinct and only become clear

with the passage of time. Control involving trust and control

through an agency theoretic approach are, therefore, alter-

natives. It is clear at which end of the continuum each applies

and, no doubt there is overlap somewhere in between where

some will prefer to trust and some will prefer to employ agency

theory, but while both approaches are o�ered to reduce the

quantity of inter-party information supplied, they are quite

distinct approaches. The consideration of when to use agency

theory and when to employ more trust is another very rich and

novel area of research for accountants with potential for dif-

ferentiating our approaches to di�erent areas of accounting

and it probably even warrants, ®rst, a separate conceptual

paper to explore the matter in more depth.

166 C. Tomkins / Accounting, Organizations and Society 26 (2001) 161±191

further disadvantages in terms of inhibiting crea-tivity and developing new capabilities.11

Wicks et al. say, therefore, that there is aninverse relationship between the willingness totrust and the need for information. This is con-sistent with the argument provided so far in thisarticle about trust being an alternative to infor-mation, with Luhmann (1979) and is also analo-gous to more recent arguments by Ring and Vande Ven (1994), Madhok (1995), Zaheer and Ven-tratraman (1995), Creed and Miles (1996) Aulakh,Kotabe and Sohay (1997) and Faulkner (1999) allof whom o�er di�erent perspectives on the trade-o� between trust and control mechanisms. How-ever, as will be argued in the next section of thispaper, the relationship between trust and infor-mation requirements are far more complex thanidenti®ed by Wicks et al. and that while, to someextent, information and trust must be alternatives,the relationship between them is far from being asimple inverse one.Wicks et al. then specify how businesses can

assess the optimal degree of trust needed in eachrelationship. Essentially they argue that the levelof trust must be matched with the level of eco-nomic interdependence between organisations.The more economically interdependent organisa-tions are, the heavier the costs of the collapse ofthat association will be and so the risks faced willbe higher. Hence, they argue that the greater theeconomic interdependence, the more one shouldbuild a closer (trusting) relationship to try to ensurethat collapse in the relationship is avoided. Wickset al. then continue to state that they would expectbetter corporate performance to be associated witha better match between interdependence and trust.There is inconsistency in the Wicks et al.'s

(1999) analysis. If information and trust are sub-stitutable uncertainty absorbing mechanisms, it isclear that ®rms that are highly economicallyinterdependent could strive to avoid collapseeither by building higher levels of trust or bybuilding more extensive control mechanisms withthe associated increase in information. It is the

level of uncertainty absorption e�ort that has tobe related to economic interdependence, not trustper se.12

It may also be argued that the very notion ofoptimal trust does not recognise that trust canexist without a calculative frame of mind. Cer-tainly, trust can exist without speci®c probabilisticcalculation. At the extreme it can occur onimpulse. The adoption of a calculative perspectivemay also seem to ignore di�ering personal pro-pensities to trust which obviously exist. In fact,despite their call for optimal assessment, Wicks etal. do argue that, although a rational cost±bene®tanalysis is an important foundation for trust, itprovides an insu�cient basis for a completeunderstanding of it. They even say there shouldalways be a bias towards trust:

Trust occurs because an emotional bond iscreated between people, enabling them totake `a leap of faith' that trust will be hon-oured (Wicks et al., 1999, p. 100).

They also say that it seems desirable to adopt amoral belief in the good character of the otherperson. Once again the argument is confusing. Itseems to backtrack on their stance of ®nding anoptimal balance between trust and information:they seem now to be expressing a preference fortrust over more information. Also their argumentis not su�ciently penetrating in identifying thedeeper relationships between information andtrust. From whence are these emotions and desiresto believe in people derived? Both the paper byWicks et al. (1999) and this paper attempt to con-sider the managerial problem of informationdesign given the prevailing culture and personalpropensities to take risks and cope with that con-dition by exercising trust and/or informationanalysis. Even if the cultural or personal variablesseem to be totally determining the level of trustgiven in any situation, the theory of evolutionsuggests that such trust does not arise without a

11 One might also argue that that a party who does not feel

su�ciently trusted might act more creatively to try to impress,

at least for a while, but that does not refute the basic idea.

12 Moreover, the choice between the information and trust

mechanisms will be both personally, culturally and situation-

ally dependent: the concept of optimality would probably need

to have a highly localised interpretation.

C. Tomkins / Accounting, Organizations and Society 26 (2001) 161±191 167

reason, even if it was never thought out explicitlyby a human mind. A person may act immediatelyand spontaneously in a trusting manner, but thatis because it is a bene®cial response learned fromprevious ``experience''.13 Furthermore, after dis-cussing the emotional drivers of trust, Wicks et al.themselves revert back to o�er a rational cost±bene®t basis for supporting a bias towards. Theyrefer to Jones (1995) who argued that, withoutsome emotional attachment and self-restraint,either opportunism by one party will occur orthere will be the need for high agency costs toprevent it. Trust that self-restraint will occur,therefore, reduces agency costs. This, however, isreally no more than a restatement of the supposedinverse relationship between trust and informa-tion. Wicks et al. provide a stronger case for morerather than less trust when they follow Ghoshaland Moran (1996) who argue that Williamson'stransactions costs economics is likely to result in a:

spiralling of distrust, where both parties in arelationship erect protections to guard againstopportunism, even though both would bebetter o� . . if they could create more trust.

But this is really saying that information pro-cessing costs not only increase as trust declines,but the inverse functional relationship betweentrust and information costs becomes increasinglyand then excessively steep as trust declines. This is

still a rational cost-bene®t argument; it is not anargument based on human emotion per se.The only empirical support o�ered by Wicks et

al. (1999) for a desired bias towards trust is Axel-rod's (1984) well known tests of repeated plays ofthe Prisoner's Dilemma and similar conclusions byFrank (1988). But Axelrod's conclusions are basedon rational cost±bene®t calculation, not emotion.This, however, matters little. Irrespective of whe-ther or not there is advantage in a bias to trust14

and why that occurs, the key point is that theremust be, to some extent, a trade-o� between trustand information need. The use of the words ``tosome extent'' may seem surprising given the de®-nition of trust set down earlier which implies thatthe more that trust is given the less information isrequired, but the use of these words was deliberatebecause the main thesis of this paper is that thereis not, as Wicks et al. or even Luhmann (1979)would have us believe, a simple inverse relation-ship between trust and information. It will beshown that the relationship between trust andinformation need is more complex and, therefore,®nding a balance between trust and information isalso more complex than Wicks et al. suggest.

2.4. Beyond a simple inverse relationship betweentrust and information need and the dynamics ofbuilding trusting relationships

It is probably true to say that, at any speci®cpoint of time, when taking any one decision, thereis a reasonably strict inverse association betweeninformation and the level of trust (i.e. trust inten-sity), but this assumes that the level of trust inten-sity is somehow given quite independent of anyother activity. It ignores the fact that trust derivesfrom learned, usually interactive, experiences andthat process itself depends upon information aswell as appropriate information depending uponthe state of trust. The adoption of an inverse rela-tionship between trust intensity and informationre¯ects a simplistic static analysis that ignores the

13 That `experience' may be transmitted genetically as in the

case of babes and their mothers or as demonstrated through

di�erences in personal propensities to trust. The cost±bene®t

analysis may then be thought of as having been carried out at

an earlier stage of the evolutionary process itself. There may

also be a more ready willingness to trust in some cultural con-

texts, but the culture will have developed from a pre-history in

which it became clear, probably without speci®c human calcu-

lation, that bene®ts accrued from such a culture. The nature of

physical and cultural evolution is such that while behaviour

may have been the result of unplanned variation, there is a

mechanism that compares di�erent forms of behaviour and

selects the most appropriate for its context. There is, therefore,

a rationality for that behaviour even if the behaviour was not

the result of a human rational decision-process. Hence one

ought not to argue in favour of preferring a bias to trust based

just upon emotion, without implying, perhaps quite deep down,

a underlying learned rationality.

14 In fact, the earlier footnote that contrasted agency theory

and trust approaches to information reduction suggest that the

use of greater trust is contingent and not universally applicable.

168 C. Tomkins / Accounting, Organizations and Society 26 (2001) 161±191

dynamic process of building trust and the role thatinformation has in that. Close relationships arenot usually created overnight Ð they take time tobuild and are expected by the parties involved toexist for some duration of time Ð that is beyondone transaction or meeting even if the duration isexpected to be limited to one speci®c project orventure. There is, in other words, an expectationof continuity and repeated meeting15. In somecases the movement to a close relationship may bequite rapid, but it is normally created by repeatedinteraction that involves the passing of informationbetween the parties concerned.16 Wicks et al. dorecognise that trust is a dynamic and continuingvariable:

Relationships unfold so that individuals con-tinually update their information base and theirdecisions to trust. (Wicks et al., 1999, p. 101),

but they do not follow through the implicationsof this.17 They also recognise that trust is con-textually dependent on cultural and personal vari-ables. My thesis, however, is that this contextualin¯uence not only involves national and organisa-tional cultures and personal propensities to trust,but also the stage of progression of the relationship.Putting relationship dynamics and contextualin¯uences of this sort together o�ers much moreinsight into how trust intensity actually interactswith regard to information needs.18 The questionof di�ering trust/information mixes for di�erentnational cultures will not be addressed here, butthe theme of information and trust needs at dif-ferent stages of developing a relationship will bepursued. It will be argued that the functionalassociation between trust and information is morelikely to be characterised over the life cycle of arelationship by an inverted U-shape, as shown in

15 This was, in fact, the basis for rejecting Wildavsky's

(1964) stark advocate/guardian explanation of public sector

budgeting (see Rosenberg and Tomkins, 1983).16 At the level of personal interaction, one can immediately

think of exceptions. For example, one may never have met a

medical surgeon before, but be willing to give trust in him/her

and his/her expertise. One might, however, argue equally that

this situation arose through learning in society over time. This

sort of trust still evolved through a process of multiple experi-

ences of how medical knowledge was expanding and how the

profession could be relied upon. There had to be experience

followed by re¯exivity. Schemas of trust can be learned from

our parents, social contacts and even books, rather than direct

experience. It can also be institutionalised in acceptance of

professional judgements. Nevertheless, information and

knowledge played a central part in establishing that trust even

if, for a speci®c individual, trust is immediately given without

question. Moreover, such schemas can be quickly broken if

further information suggests the trust is misplaced and there

has clearly been a questioning of the standing of some profes-

sions and professionals in recent years. It is also important to

address Meyerson, Weick and Kramer (1996) and their analysis

of ``swift trust''. They examine the increasing phenomenon of

organisations coming together quickly in temporary systems in

such a swift manner that detailed prior analysis is impossible.

At ®rst sight this refutes the argument of this paper which

suggests that trust is a longer-run learned activity. These

authors conclude, however, that swift trust is less about close

relations than agreeing to take action based on a few generic

features of the setting and people involved. Organisations

involved in temporary systems also embarked on a signi®cant

degree of hedging of risks. This can be seen as nothing more

than an initial stage of a movement towards a long-term rela-

tionship (as described in the next section of this paper) that

never fully matures. Moreover, swift trust may be made possi-

ble by established notions of trust attached to social positions

as just discussed. Yet furthermore, in one of the very few

studies to date relating trust to ®nancial analysis, Harrison,

Dibben and Mason (1997) show that while new ventures often

gain support on the basis of just broad screening, there is a

heavy reliance of business angels upon the advice of entrepre-

neurs who have a longer run relationship with both the fund

seeker and fund provider and who create the basis of the rela-

tionship. Hence, trust built up over the longer term with third

parties can facilitate their contacts to enter into trusting rela-

tionships much more quickly, but, somewhere in the network,

trust had to be built through stages. The swift trust phenom-

enon does not invalidate the proposals in this paper.17 They were, of course, not so directly concerned with this

as with the relationship between success, inter-dependence and

trust.18 Lewicki and Bunker (1996) also emphasise the dynamics

of trust relationships and argue that trust develops through

consecutive stages of Calculus-based trust (market orientated,

economic calculation), Knowledge-based trust and then

Stable-identi®cation-based trust). As Harrison et al. (1997)

say, such a framework ignores many other factors which a�ect

trust. The analysis in the current paper tries to focus more

comprehensively on information needed for all types of factors

which may lead to trust. Moreover, it argues that there is a

need for some Calculus-based trust at all stages of trust

development as illustrated by the introduction of two infor-

mation types: Type 1 and Type 2 (Calculus-based trust) dis-

cussed later in the paper.

C. Tomkins / Accounting, Organizations and Society 26 (2001) 161±191 169

Fig. 1, rather than a monotonic inverse associa-tion between trust and information.19

Through repeated interaction, a closer relation-ship with increasing trust may develop success-fully. This may reduce the risk that the activityassociated with the relationship will not collapseand the need for information will then be reducedgiving the initial argument for an inverse associa-tion between trust and information. However, atan early stage of a relationship, commitments areusually less extensive and so the risks attached to

breach of that commitment are lower than at laterstages of the relationship. Hence, there will be alower need for either trust20 or information as anuncertainty absorbing mechanism. Moreover, it isposited that, as the relationship matures from theinitial state of low level of trust, there will be apositive association between trust and informationsimply because trust itself cannot be increasedwithout further information. So, once one takes adynamic view of a relationship, one sees that there islikely to be a positive association between informa-tion and trust at earlier stages of relationship devel-opment. As trust intensity becomes established athigher levels in later stages of the relationship, it islikely that less information will be needed to sustainthat relationship. Hence a negative association

19 It is argued that there will be an inverse U-shaped func-

tion in all cultures, but that the shape of the function may well

di�er because of cultural propensities. In Japan, for example,

the upswing of the inverted U may be steeper through the pro-

pensity to place more reliance to build consensus prior to

making a decision and then relying more on trust and less on

detailed accounting to monitor subsequent actions [See the

analysis of Japanese accounting in The Ford of Europe (A)

case, IMEDE Case GM-321, (1984) for a description of sim-

pli®ed monitoring procedures used by Japanese motor manu-

facturers].

Fig. 1. The inverse u-shaped relationship between trust and information. (A reviewer suggested that the curve might be more like a

nested or stepped series of curves with increases in slope as one moves to higher levels of trust. She/he may be right, but enough

argument is needed to establish the general U-shape in the ®rst place before one can get too re®ned).

20 In fact, it would probably be correct to say that the issue

of whether to trust does not even arise at early stages of a new

relationship. A low level of trust in this sense does not mean

distrust; it indicates that less signi®cant consequences are

attached to a break-down in the relationship.

170 C. Tomkins / Accounting, Organizations and Society 26 (2001) 161±191

arises at later stages of building trust. These twoe�ects give rise to the inverse U-shaped functionrelating aggregate information needs to trustintensity and the maturity of the relationship. Itseems important to grasp this in consideringinformation system design. While requests forexcessive information may reduce trust, the provi-sion of more information is needed to create moretrust in the ``upswing'' of the relationship Ðinformation in¯uences trust which in¯uencesinformation need which in¯uences trust and so on. . . This is a rather more complex associationbetween trust and information than Wicks et al. oreven Luhmann suggest.

2.5. Information needed to support trustdistinguished from information for mastery ofevents

In fact, the argument cannot rest there Ð theassociation between trust and information is evenmore complex. Luhmann (1979) stresses the dif-ference between information needed when trust isabsent and information for the `mastery of events'.The recognition of the need for information tomaster events is important. Whatever the level oftrust in a relationship, information will still beneeded to help plan a collaborative future by set-ting down what each party wishes to achieve fromthe collaboration, how feasible the goals and rela-tive roles are and what actions need to be taken(as, for example, described in the interactivesimulations earlier in this paper). Informationneeded in the absence of trust is more related tothe veri®cation of actions of the other party inaccordance with expectations rather than what theparties should collectively do. While Luhmann isright that one can never dispense with trust com-pletely because one can never be omniscient,equally one can never dispense completely withthe need for some information on the other party'sactions unless trust is absolute in a blind andunthinking manner. Even so there is a need tobroaden Luhmann's concept of information in theabsence or reduction of trust even further. It isnecessary to recognise the dynamics discussedearlier to incorporate information needed to helpcreate, rather than just monitor, trust.

It is suggested, therefore, that distinguishingbetween two main types of information needs willhelp any analysis of trust and information. Type 1is information that relates to the willingness totrust; that is what is needed to create trust andcheck on the state of the relationship. Type 2 isinformation needed for mastery of events by thatrelationship as an entity itself. It will be suggestedbelow that information Type 1 and informationType 2 will both have an inverse U-shaped func-tion with respect to relationship maturity.21

Moreover, while it is obvious that InformationType 1 will in¯uence how trust intensity developsand thereby feed back into the future requirementfor Information Type 1, aspects of InformationType 2 will also in¯uence trust intensity even ifthat type of information was not prescribed spe-ci®cally to assess the degree to which one cantrust a partner. So these two types of informationare not totally separate from each other. Colla-borative involvement in the process of planningthe mastery of events as described earlier in thispaper can itself help to create or destroy trust,but, as will be shown below, there is a di�erentemphasis in content to the information neededmainly to plan and make decisions on collabora-tive futures compared to that needed mainly forbuilding trust in the actions of ones partners.Type 1 information relates to information on

21 Das and Teng (1998) also conclude that a simple sub-

stitutive relationship between trust and control mechanisms

restricts our understanding of how con®dence in another's

actions is gained. They say that high levels of trust will make

control mechanisms more e�ective. This is not, however, the

same argument related to the dynamics of the interaction

between trust and information used in this paper. The logic

used by Das and Teng leads them to state (their Table 2, p. 500)

that one can only have a high degree of con®dence in a partner

where both the control level and the level of trust are high. This

seems to be too extreme a position, while supporting Das and

Teng's call to question simplistic substitutability between trust

and control mechanisms, there must be some degree of sub-

stitutability or trust seems to lose much of its meaning. Das and

Teng (in press) do, however, specify di�erent types of trust

(performance trust, competence trust, etc.) and how certain

types of control and trust are more important than others in

particular alliance types. There would be mileage in examining

more closely in future work the relationship between their pro-

posals and the detailed information listings in this paper.

C. Tomkins / Accounting, Organizations and Society 26 (2001) 161±191 171

competence and on integrity in both action andcommunication [see, for example, Reina andReina (1999) and Das and Teng (in press), fordiscussions of contractual trust, communicationstrust, goodwill trust and competence trust].Information Type 2 concerns planning what eachparty is going to do.There is one ®nal dimension to consider in

looking at the association between trust andinformation. System designers cannot just focusupon how much or what content of information tosupply to match the current state of trust intensity,but also must be concerned with the technicalquality and reliability of information itself. Dis-trust of an individual to supply full and accurateinformation is the same as a lack of communica-tions trust based upon perceptions of reputation,professional a�liation, track record, etc. There isalso the question of how feasible it is technically toprovide a clear and accurate picture even wherethere are good intentions. The rest of the paperwill assume that technically reliable informationcan be provided. In practice, of course, accountingis greatly concerned with the technically mostreliable forms of measurement, materiality, etc. Insituations where matters are technically di�cult tomeasure, relationship management through trustbecomes more important.We are now in a position to consider the prac-

tical implications of such a theoretical basis on thedesign of information needs in, ®rst, personalrelationships and, second, business relationships.

3. How trust and information needs change indeveloping personal relationships

Ford, Gadde et al. (1998) stress four di�erentstages of development in buyer±seller relation-ships: the pre-relationship stage, the exploratorystage, the developing stage and the stable stage.Similar stages are also evident in the developmentof personal relationships. Moreover, we all haveexperience of the development of such relation-ships and so reasoning from this base should bequite meaningful to each of us because we all havemuch internalised research data to contribute onthis question.

Belief that a consideration of personal relation-ships22 would provide some insight into the infor-mation needs of businesses entering alliances ledto the development of Table 1. That table wasconstructed by re¯ecting upon the theoreticalpropositions set out above regarding the way inwhich dependence, trust and information needschange as the personal relationship develops andthe distinction between Information Type 1 (nee-ded for willingness to trust) and Information Type2 (for mastery of events). An e�ort was then madeto think through more speci®cally what kind ofinformation was needed at each stage of the rela-tionship for both information categories.The model in Table 1 is ``idealised'' in the sense

that it assumes, like the Ford et al. business model,that the relationship moves progressively throughfrom an initial phase to maturity and stability. Ofcourse, many personal relationships do not andmany go into reverse progression and, in responseto events, distrust can follow trust. In fact, evenmost lasting relationships ¯uctuate to some degreeeven if that is around a central trend towardsincreased trust intensity. Also, movement backdown the `trust development chain', perhaps dueto a breach of trust, may call for a signi®cantlydi�erent mix of trust and information than wasneeded on the way up. Hence, Table 1 is o�eredonly as a rough mapping of the trust/informationmix applicable at each stage of such an `idealised'development.The chief features of the model shown in Table 1

and associated comments are as follows:

22 Of course, there are all sorts of personal relationships, but

I chose to address this just in the form of long-run partnerships

between two people leading to co-habitation and building a

family in the belief that this would provide a clearer insight into

di�erent relationship stages. This is usually where inter-

dependence is highest in our personal relationships. The devel-

opment of lasting and deep friendships probably goes through

a similar process. More casual relationships may not, but they

do not involve high interdependence. Also each individual

probably does not got through such an extended process in the

development of trust in professional advice or a teacher, but it

was argued earlier that this is because society has, over time,

taken those steps for the individual through a process of insti-

tutionalisation of trust.

172 C. Tomkins / Accounting, Organizations and Society 26 (2001) 161±191

Table 1

An `idealised' life cycle of trust/information needs for developing personal partner relationships

a. Information type 1: needed to assess willingness to trust

Degree of

inter-dependence

Degree of commitment

(i.e. trust intensity)

Questions to be

answered

Information Type I Volume of Information Type I

needed and how communicated

1. Low Exploratory relationship,

Minimal commitment

and trust

What do I need to

know to ®x a `date'

and further `dates'?

Basic personal attributes

(depending upon type of

`date', etc.)

Understanding of personal

values

Limited, general, impressionistic.

2. Prepared to

increase

Increasing trust:

short run commitment

What do I need to now

about partner before I

get more involved?

How much should I

reveal of myself?

Aspirations / expectations

Adaptability shown to me and

expected from me

Degree of interest in me

Formation of ``®rewalls''

Frequent consideration

of behaviour, but guarded,

statements of wishes, capacities,

intents and facts.

3. High High trust: long term

Commitment made and

regularly con®rmed

What do I need to

con®rm my

continuing trust?

Actions reliably undertaken

according to anticipated values

Commitment given to me rather

than others.

Implementation of translucent

information systems

Quite extensive and open

Largely based on frequent

informal communication and

observation of behaviour

4. Very high High trust: stable

commitment for

remaining life

What does my

partner need to

con®rm continuing

trust?

Mainly ``exception signals''

Presumption of reliability based on

broad results

Transparency: activities available

for inspection

Limited, informal, general

indicators of satisfactory

relationship

b. Information type 2: needed to attempt a collaborative mastery of events

Degree of

inter-dependence

Degree of commitment

(i.e. trust intensity)

Questions to be

answered

Information Type 2 Volume of Information Type 2

needed and how communicated

1. Low Exploratory relationship,

Minimal commitment

and trust

Is relationship worth

immediate e�ort?

Short term cost±bene®t

Likelihood of ``option value''

derivable from the relationship

Low: broad comparison of

obvious alternatives and

consideration of other

possibilities, extensive rational

analysis not necessary

2. Prepared to

increase

Increasing trust:

short run commitment

If I get involved,

what economic and

other implications

does this have?

Assessment of future jointly

compared to separately

Tentative scenario planning,

broad consensus over paths to

follow

More extensive; comparison with

other possible relationships

initially impressionistic moving

to more rational, but broad,

strategic analysis

(continued on next page)

C.Tomkins/Acco

unting

,Organiza

tionsandSociety

26(2001)161±191

173

. Relationships are built through stages ofexperience and trust intensity. Inter-dependence and information needed by eachparty varies according to the stage of therelationship.

. A relationship will not develop without thegrowth of trust, but willingness to continuewith the relationship depends upon morethan the existence of trust. The relationshipmust deliver something of value to the par-ties involved as well as being simply an arenafor trust. It is, therefore, a mistake to put allinformation needed to build and sustain arelationship into the same category. It makessense to make the distinction between Infor-mation Type 1 (needed to indicate will-ingness to trust) and Information Type 2(needed to plan a collaborative existence andmastery of events). There are di�erences incontent and emphasis in the kind of infor-mation needed for each purpose even thougheach type can contribute, to some extent, toboth purposes.

. The questions that need to be answered to (i)establish willingness to trust and (ii) to plan acollaborative mastery of events, both changewith the development of the relationship.This is due wholly to changes in trust inten-sity in the case of Information Type 1.Information Type 2 needed partly dependson the degree of trust (e.g. limiting oneself toscenario planning in stage 2 when the rela-tionship is still building trust), but moreimportantly upon the type and scope of theeconomic decisions to be made at each stageof the relationship.

. The volume of detailed information for bothinformation types required increases andthen decreases with increases in trust inten-sity as the relationship matures. This seemsto bear out the inverted U-shaped functionbetween relationship maturity and informa-tion need as hypothesised above.

. The model is ``idealised'' in the sense of mov-ing steadily through to maturity, but also inthe choice made by the relationship parties asto mode of operation at each stage. The pat-tern o�ered is expected to approximate theT

able

1(continued)

b.Inform

ationtype2:needed

toattem

ptacollaborativemasteryofevents

Degreeof

inter-dependence

Degreeofcommitment

(i.e.trust

intensity)

Questionsto

be

answ

ered

Inform

ationType2

Volumeofinform

ationType2

needed

andhow

communicated

3.High

Hightrust:longterm

Commitmentmadeand

regularlycon®rm

ed

Whatdoweneedto

investin

tobuildour

future?

How

should

we

allocate

jointresources?

Choices

over

housing,career

orientation,leisure

activities,

educationofchildren,saving

forfuture

Balancingtimebetween

competingdem

ands

Clearexpectationsofeach

other

Detailed

resourceallocation

analysesbasedonform

al

inform

ation(perhapsdependent

onwealth)andform

allycon®rm

ed.

Monitoringeach

other

against

expectations

4.Veryhigh

Hightrust:stable

commitmentfor

rest

oflife

How

canweenjoy

whatwealthandtime

wehaveleft?

Inform

ationonsharingtimeand

new

experiencesto

keep

relationship

fresh,

Oversightofinvestm

ent

portfolio,endowments

Low

levelsofform

alanalysis

(dependentonwealth)

174 C. Tomkins / Accounting, Organizations and Society 26 (2001) 161±191

usual mode,23 but those involved can alwayschoose to operate di�erently. Choice24 isexercised and the process has to be managedand so it may always be possible to ®nd somerelationships deemed acceptable by the par-ties involved that do not conform to thearchetypes in Table 1. The model is also`idealised' in the sense that it suggests fourdistinct stages of development. This is merelyto facilitate description and discussion.Movement through the apparent stages maybe more like a continuum of ever changingtrust intensity or, conversely discontinuouswith leaps between levels of trust.

. The volume of detail required for Informa-tion 2 will be dependent on the wealth posi-tion of the partners in the relationship relativeto their personal aspirations Ð especially instages 2, 3 and 4. People with high wealth,relative to aspirations, can plan relativelyexpensive futures without detailed analysis instages 2 and 3, while poorer people (or thosewith high spending aspirations compared totheir wealth level) still have to ``count thepennies'' in stage 4 to remain solvent.

. The type of information, degree of detail andmethod of communication has a di�erentemphasis in di�erent stages and itself engen-ders movement to higher (or lower) levels of

trust intensity and, hence, di�erent informa-tion needs. The appropriate matching ofinformation to trust intensity itself changesthe information need in a dynamic process.

. The information problem addresses whateach party needs for their own current level oftrust intensity, but also what each thinks he/she has to give to meet what the other partyneeds for their level of trust intensity. Notethat what each party wants and what thepartner thinks he/she needs are not necessarilythe same and some further information ¯owmay be needed to sustain a reasonable level ofconsistency between these two assessments. Itshould also be noted that each party's trustintensity at any time may be quite di�erentfrom that of their partner and, hence, eachparty's information needs may be di�erent.Moreover, each party may have a di�erentpersonal propensity to trust. This observa-tion also suggests that if each party wishes tomaintain the relationship, he or she mustbear the cost of giving information that he orshe may not themselves require. Radical dif-ferences in trust intensity are likely to resultin a breakdown of the relationship, but itmay not if the cost±bene®t analysis under-taken using Information Type 2 suggests thatthe less trusting party is still better o� withinthis relationship rather than another. At thispoint it would be appropriate to introducethe concept of the distribution of powerbetween parties to the relationship, but, asalready stated elsewhere, the analysis in thispaper is already quite complex. If this analy-sis based on an equal distribution of powermeets with a degree of agreement, the powerdimension can be added later.

4. Is interpersonal trust the same as inter-orga-nisational trust?

One ought not to move straight from an analysisrelevant for personal relationships like Table 1 toapply that structure to inter-organisation rela-tionships. One should ®rst, consider whether inter-organisational trust is su�ciently like personal

23 The word `mode' is used here in both its general and

mathematical sense. This is mentioned because Wicks et al. set

up their search for an optimal level of trust upon the founda-

tion of Aristotle's `golden mean', which they de®ne as a state

between excess and de®ciency in human conduct. According to

Urmson (1988, Chapter 2), Aristotle's `golden mean' is not a

tendency to the mathematical average, but depends upon a

judgement as to the appropriate matching of behaviour in any

given situation Ð e.g. what would be excessive emotion in one

situation can be appropriate or even de®cient in another. The

position adopted in this paper is not Aristotleian in the sense

that the model in Table 1 is held to be a general mode of

behaviour which will be adopted by a large section of the

population and, hence, does constitute a mathematical concept,

but the model is Aristotleian in the sense that it seeks to ®nd

contingent information for level of trust intensity as did Wicks

et al. in a much more general way.24 Choice here is really the outcome of an interaction rather

than a unilateral choice. If it is the latter, continuity of the

relationship itself can be at risk, although this does not neces-

sarily follow.

C. Tomkins / Accounting, Organizations and Society 26 (2001) 161±191 175

trust for appropriate parallels to be drawn. Thereare several issues that need to be addressed:

(a) Can an organisation exhibit trust as a per-son can?

(b) People change organisational positions fairlyfrequently. How does that a�ect trustbetween di�erent organisations?

(c) People work in hierarchical structures inorganisations? How does this a�ect trust?

(d) Inter-organisational arrangements will cer-tainly be subject to a contract of some type,doesn't that fundamentally reduce the needfor trust in inter-organisational relation-ships?

Point (a) raises the issue of whether an entitylike an organisation can ever be said to be actingin a trusting manner. It is axiomatic that trust canonly be exhibited by a living being. One might,nevertheless, think of an organisation acting as ifit exhibits trust. An organisation may be thoughtof as a number of people who agree to act as agroup. Clearly that group can then agree to placetrust in certain things, people or other groups. Inthis quite practical sense, organisations can havetrust in each other.25 If one can talk sensibly aboutan organisation having a culture, one can talk as ifan organisation has a degree of trust intensitytowards di�erent other organisations.This also partly answers point (b). Any con-

troller using something like Table 1 to review thestate of information given the current stage of aninter-organisational trust relationship would haveto make allowances for changing personnel andthat might mean some sort of iterative movementbetween di�erent stages of trust intensity andinformation provided.26 Even so, some continuity

of organisational trust intensity might be assumedas newly appointed individuals are conditioned toaccept a prevailing organisational schema of trustfor another organisation. As with all forms oftrust, the schema will be continually tested andnewly appointed people may look to test theorganisational schema more than well-establishedemployees. Such initial scepticism may lead to abackward iteration in the trust relationship asindicated above, but it could also mean that a newindividual in an organisational relationship maymovemore rapidly to higher levels of trust than if heor she were in a personal relationship (McKnight,Cummings & Chervany, 1998). That will not, how-ever, invalidate the notion of an organisation tak-ing time to go through di�erent stages to buildtrust. While a new employee may immediatelyaccept a current advanced level of trust betweenher/his company and another organisation, therewill be a life cycle to the inter-organisational rela-tionship itself. So it is proposed that the staged lifecycle model is still relevant for the analysis ofinter-organisational trust and information ¯ows.In contrast, point (c) may present a di�erent

aspect compared to personal relationships. Whenan individual trusts another in a personal rela-tionship, the consequences of misplaced trust pre-dominantly a�ect just that person and, perhaps,close associates. If an executive misplaces trust inanother company and its personnel, he will haveto account to his superiors and the owners of thecompany. These extra layers of accountabilitymay well persuade the executive involved in analliance with another organisation to be more riskaverse than he/she would otherwise be and insiston more or di�erent information for any givenlevel of trust. Also superiors or owners, who are at``arms length'', may also insist on more formalinformation about the state of that relationship,but, over time, such superiors and owners maywell take the lack of problems as a sign that theyneed not worry and, hence, need less information.In which case the general functional associationbetween trust and information will still holds.Points (a), (b) and (c) do not seem to invalidate thegeneral notions of building inter-organisationaltrust over time and the generic consequences forinformation requirement as that trust develops as

25 This may, as suggested elsewhere in this paper, be facilitated

if the two organisations have similar internal and external cul-

tures, but that is not a necessary condition. Organisations from

quite di�erent cultures may build quite trusting relationships just

as, with tolerance and adaptability, people from quite di�erent

cultures or backgrounds may build successful marriages.26 This, of course, also happens in relationships between two

individuals, but this takes place by a modi®cation to the trust

knowledge base of one or both persons Ð it not the same as an

entirely new person taking responsibility for an existing inter-

organisational relationship.

176 C. Tomkins / Accounting, Organizations and Society 26 (2001) 161±191

outlined above. Point (d), concerned with howcontracts a�ect trust, needs rather more con-sideration.The existence of contracts might be seen to give

rise to two problems in applying the trust/infor-mation life cycle analysis to organisations. First,contracts might be seen to exist directly becausetrust is missing. Parties do not, however, contractwith each other until they have reached a givenlevel of trust (i.e. where they are prepared tocommit themselves to each other). That did notoccur until the third level of trust in the simpleclassi®cation given in Tables 1 and so contractscannot render the earlier stages of building trustirrelevant. Even so, does trust become redundantonce the contract has been written? It does not. Itwas stressed earlier that it is impossible to removeall uncertainty whatever amount of informationone has. The same must be true of contracting Ðone can never write clauses to cover every even-tuality. Indeed, the major value of contracting maylie in it being a device to set down goals and meth-ods to enable e�ective mutual planning, ratherthan being primarily an ex-post control device. Infact, contracting may not even do that very welland there is evidence to support this view.In what is now considered a classic paper on the

use of contracts, Macaulay (1963) reported theresults of a wide-ranging empirical analysis andconcluded:

Businessmen desire to `keep it simple andavoid red tape' even where large amounts ofmoney and signi®cant risks are involved. . .and

. . . that while detailed planning and legalsanctions play a signi®cant role in someexchanges between businesses, in many busi-ness exchanges their role is small.

Macaulay explains this by stating that it seemsthat businessmen often do not want to resort tothe law to resolve non-performance issues. Theyare often aware that they may have to deal withtheir partner again or that, if they get a reputationfor ``sticking to the letter of the law'', they may

®nd it more di�cult to ®nd new collaborators.Also, if they are too strict on the other partyobserving every detail in the contract, they willlose their own ¯exibility to modify actions as theenvironment shifts.Macaulay conducted his research in the early

1960s and it is general knowledge that, since then,there has been an increasing tendency to resort tolitigation in society (especially American society),but these views were still seen by Macaulay to beappropriate in the late 1970s (Macaulay, 1977)and in 1985 when he conducted a review of his1963 study. Macaulay (1985) stated:

(Law) writers assume a number of things aboutthe institution of contract. First, there is carefulplanning of relationships in light of legalrequirements and the possibilities of non-per-formance. We must spell out everythingbecause parties will perform only to the letterof a contract, if they go that far. Second,contract law is a body of clear rules so that itcan facilitate planning. It provides formalchannels so that we know the right way to pro-ceed to produce desired legal consequences.Finally, contract litigation is a primary meansof determining breach and directly and indir-ectly resolving disputes. Without contract lawand the State's monopoly of the legitimateuse of force, performance of contracts wouldbe highly uncertain. However, all of theseassumptions about history and about humanrelationships are just wrong or so greatly over-stated as to be seriously misleading. Contractplanning and contract law, at best, stand atthe margin of important long-term continuingbusiness relations. (italics added).

Moreover, Macaulay's work does not exhibit apeculiarly American perspective. Beale and Dug-dale (1975) studied engineering manufacturers inthe UK and found very similar results to thosefound by Macaulay. While Vincent-Jones (1985)was able to conclude:

. . . whilst the legal institution of contractundoubtedly plays a central role in certain

C. Tomkins / Accounting, Organizations and Society 26 (2001) 161±191 177

circumstances,27 it is clear that the `extra-contractual' constraints of convention, cus-tom and expedience are of far greater andmore direct importance in the vast majority oftransactions (italics in the original).

One does not have to go half so far as Macaulayor Vincent-Jones (1989) to justify the conclusionthat the existence of contracts in inter-organisa-tional relationships leaves plenty of space for thereduction of uncertainty through building trustand that this will need to be supported by theprovision of information much as outlined in thispaper. Also, with quite di�erent arguments fromthose of Macaulay and Vincent-Jones, Quinn andSpreitzer (1997) state that trust is not just analternative to contracts, it adds value to the con-tracted relationship by encouraging a sharing ofinformation and resources. It, therefore, makessense to look for a parallel to Table 1, which mightapply to inter-organisational trust Ð the existenceof business contracts do not render this irrelevant.

5. Trust and information in business relationships

It now seems almost self-evident, that featuresof information in business relationships can beidenti®ed which are analogous to the eight fea-tures of the model in Table 1. All relationships aredynamic and can develop in intensity and mayevolve through stages similar to those shown in

Table 1. It also seems obvious that the sameconceptual distinction can be made between Infor-mation Type 1, needed to facilitate trust, and Infor-mation Type 2, needed to make economicjudgements on strategies, investments and on-goingoperations. So, using Table 1 as a template, Table 2was constructed to suggest a set of propositionsrelated to the information characteristics that onewould expect to ®nd at di�erent stages of develop-ment of trust intensity in a business relationship.If these propositions do have validity, it would

be appropriate for a company to assess exactlywhat type of information it needed to satisfy thecharacteristics shown in Table 2 and ensure that itwas available at the right time in each relationship.The need to recognise timing is a key ®nding andis crucially important because one should notexpect too much from the relationship too soon.The information received in one stage determineswhether there will be progression to a higher stateof interdependence and trust intensity. Thisemphasises that designers of business informationsystems should not be seeking a given time-invar-iant optimal system, but they have a responsibilityfor perpetually adapting and modifying the systemto meet changing relationship needs. Equallyaccountants should consider how they can con-tribute to the speci®c information needs listed inTable 2 and assess at what stage in relationshipbuilding the di�erent aspects of accounting arelikely to be acceptable without destroying trust. Itwas stated at the beginning of the paper thatoverheads in one organisation might drive over-head expenditure in another. The question willthen arise as to whether to probe these inter-orga-nisation e�ects because they are vital for masteryof events or whether to accept some fuzziness onsuch issues because one is at the stage where `®re-walls' are still necessary or, alternatively, that onedoes not want to demonstrate a lack of trust at asensitive time in building trust and, perhaps, moveinto Ghoshal and Moran's spiral of distrust. Thefundamentals of accounting techniques will notchange, but the timing of introducing them at dif-ferent stages of the relationship seems important.Accounting techniques and processes must not beintroduced just because they seem logically relevantfor mastery of events, even though some

27 One should take care not to assume general tendencies are

universal rules. For example, the UK construction industry

seems to place an excessive emphasis on control through con-

tracts (as discussed in Nicolini et al., in press). In fact, contracts

are widely used in that industry with the likelihood of con¯ict

in mind. This does not undermine the arguments in this paper.

Clearly there will be some contextual variability in the extent to

which one can rely on the ®ndings of Macaulay, Beale and

Dugdale. The UK construction industry seems to be at one

extreme of such variability and it is consistent with the argu-

ments in this paper to ®nd a greater reliance on contracts where

trust is quite low, but, even at this extreme, contracts cannot

absorb all uncertainty and this seems to be supported by the

high relative incidence of legal cases relating to construction

that are only resolved upon appeal to The House of Lords

(Rhys Jones, 1991).

178 C. Tomkins / Accounting, Organizations and Society 26 (2001) 161±191

accounting will be vital for this purpose. Trustneeds to be built with care. Lamming, Caldwelland Harrison (1999) discuss how inter-companyaccounting transfers do not exist in a near dichot-omous form with or without a ``®rewall''. Theyuse the concept of ``translucence'' to indicate amixture of opaqueness and openess. Even thatmay be too simple an interpretive structure. Rea-lity is probably more complex with more of acontinuity of exposure as a relationship deepensand, perhaps, even some revelations being moreapparent than real. The emergence from opaque-ness, through di�erent degrees of translucence tosigni®cant transparency o�ers another very richpossibility for empirical research in accountingwhich has signi®cant practical implications for theway accounting can serve the building of business

alliances. The cost±bene®t analysis of accountingsystems must incorporate the trust dimension andhow that changes with the maturity of the rela-tionship; this dimension needs to be added to ourempirical research into ``everyday accounting''.

6. Forms of business alliances and their informationneeds

The discussion so far has been couched in termsof the general features of information required toserve all relationships that deepen over time. Thisis insu�cient, however, for specifying the precisebusiness information needs for such relationshipsbecause there are di�erent forms of business alli-ances. It is not just the development of the relation-

Table 2

Hypothesised information characteristics at di�erent stages of business relationship development

Development stage Information type 1 (to warrant trust) Information type 2 (to master events collaboratively)

Exploratory/

screening

Possible partners' attributes: ®nancial,

technological, market and network

positioning based largely on publicly

available information

Values, integrity and ethics:

the way they do business

Costs and bene®ts of initial experimental ventures

Broad assessment of relationship option values

Building

commitment

More detailed (internal) exchange of

information on speci®c partner's attributes

Aspirations

Adaptability exhibited and requesteda

Establishment of relevant

information ``®rewalls''

Con®dentiality demonstrated re.

information released

Scenario development

Guarded revelation of costs and bene®ts of

strategic options

Why interdependence is required

Comparison with alternative relationships

Long term

commitment

established

Information that processes observed

Reliable achievement of milestones, costs,

quality, etc. Moving back ``®rewalls'';

development of translucency of systems

Degree of attention provided to our

relationship compared to others.

Detailed assessment of joint competitive position

Detailed investment appraisal (variations according

to type of investment)

Agreed expectations of each other

Pro®t/risk sharing schemes

Later life

in the

relationship

Open book accounting for speci®c projects

Transparency of actions / pro®ts-

available for examination if required,

but not routinely exercised

Main focus on results and lack of crises, not

process control

How to ``milk'' what we have

Possibilities for extending relationship to new

businesses, markets, technologies, etc.

a See HaÊ kansson (1982) and Brennan and Turnbull (1998) for discussion of adaptations often required in buyer±seller relationships

and how to classify and measure them.

C. Tomkins / Accounting, Organizations and Society 26 (2001) 161±191 179

Table 3

Information needs to types of alliance at di�erent stages of development

Forms of

alliance

Main purpose Exploratory/screening Building commitment Long term commitment

established

Later life

Favoured

subcontracts

Cost reduction

and quality

improvement

Information type 1

Reputation for quality,

price, delivery

Likely to stay in business

Information type 1

Quality checks on deliveries

Contracted targets for cost

reduction

Assessment of supplier R&D

Information type 1

Translucent books on

cost data

Information type 1

No checks on deliveries

Open books on cost data

Information type 2 Information type 2 Information type 2 Information type 2

Make vs. buy for short term

contracts

Comparison with alternate

suppliers

Make vs buy analysis for

extended basis on own design

Assessment of longer run

development potential

Joint e�ciency planning

Design sub-contracted

Involvement in target costing of

new products

Analyse new market

possibilities

Technology

licensing

Speed of access to

technology at

lower cost

Information type 1

Standing in on-going

technology development

Trial short term licensing

arrangements

Information type 1

Demonstrated e�ectiveness of

technology purchased in

enabling speed to market

Prospects for long run

continuity

Information type 1

Adaptiveness of supplier's

technology development to

buyer's needs

Information type 1

Common knowledge of

technology development

plans

Information type 2 Information type 2 Information type 2 Information type 2

Comparison of in-house

development time with

buying in on licence

Comparison of alternate

providers

Detailed cost±bene®t analysis

of longer term licensing vs. in-house

development (including costs of

maintaining ability to re-enter

development)

Strategic analysis of abandoning

competence in this technology

Involvement in speed to market

aspects of target costing

Analysis of new market

Possibilities

Strategic

alliances

(including

PFI consortia

in the UK)

New product

development (near

market)

Provision of end

service, rather than

capital goods

Information type 1

Identi®cation of other

organisations' competencies

in similar technologies or

products

Information type 1

Technical knowledge gained

from initial collaborations

Willingness to share risks

Con®dentiality demonstrated re.

information released

Information type 1

Adaptiveness to needs

of supply chain evident in planning

No immediate resort to

contract terms and legal

means if when problems occur

Translucent books on

costs and revenues and technology

Information type 1a

Satisfactory ®nancial

results from

collaborations

Open books on costs

and revenues and

technology

Information type 2 Information type 2 Information type 2 Information type 2

Broad assessment of ®nancial

possibilities compared to rival

consortia

Analysis of dimensions for

collaboration and likely allocation

of responsibilities

Consideration of alternate

alliances

Target costing

Joint investment appraisal and

pro®t sharing analyses

Analysis of new market

possibilities

Making a cash cow of

the project/product for

remainder of life.

180

C.Tomkins/Acco

unting

,Organiza

tionsandSociety

26(2001)161±191

Research

consortia

New technology

development

Information type 1

Perceived need to collaborate

to cope with high expenditure

and risk

Patent record

Information type 1

Technical knowledge gained

from initial collaborations

Willingness to share risks

Con®dentiality demonstrated re.

information released

Information type 1

Meeting project milestones

No immediate resort to contract

terms and legal means if problems

occur

Limited exposure on technology

Information type 1

Research patents

developing jointly

Open books on

technology

Information type 2 Information type 2 Information type 2 Information type 2

Likely risk/cost savings from

shared research

Cost sharing schemes for pilot

projects

Consideration of alternate

consortia

Contributions of research to

value chain

Investment appraisal and budgets

for speci®c programmes

Pro®t sharing schemes

Possibilities for new

research areas

Joint

venturesbNew businesses Information type 1

Partners' attributes: ®nancial,

technological, market and

network positioning based

largely on publicly available

information

Values and ethics: the way

they do businessc

Information type 1

More detailed (internal)

exchange of information on

speci®c partner's

attributes and aspirations

Adaptability exhibited and

requestedd

Establishment of relevant

``®rewalls''

Con®dentiality demonstrated re.

information released

Information type 1

Information on plans

Information that processes

observed and plans carried out

Reliable achievement of

milestones, pro®ts, costs,

quality, etc

Increasing revelation of

information on total operations

of JV to parent organisations

(especially important where JV

is by contract rather than

separate company)

Information type 1

Satisfactory JV ®nancial

results

Openness of information

on total

operations of JV to

parent organisations

Information type 2

Analysis of need for joint

venture rather than other

shorter term alliance

Information type 2

Broad strategic analysis for new

business

Why interdependence is required

Analysis of di�erent joint venture

possibilities

Information type 2

Detailed assessment of

competitive position

Detailed investment appraisal

Financial structure, equity

division and dividend policies

Information type 2

New strategic

development possibilities

a Though strategic alliances tend to be for speci®c projects or product developments for a ®nite life span.b It is not surprising that the information characteristics which appear in this part of the table are similar to that in Table 2 as the speci®cation of the purpose being the establishment of a new

business is very broad.c See Carr, Tomkins and Bayliss (1994, Chapter 5) for an example of an Anglo-German JV which had di�culties through contrasting control styles, although, as stated earlier, good rela-

tionships may be built between organisations with di�erent cultures provided partners are ¯exible.d See HaÊ kansson (1982) and Brennan and Turnbull (1998) for discussion of adaptations often required in buyer±seller relationships and how to classify and measure them.

C.Tomkins/Acco

unting

,Organiza

tionsandSociety

26(2001)161±191

181

ship, but also the speci®c purpose of the relationshipand the need to master events to achieve that pur-pose that determines information needs. Conse-quently, an attempt was made to consider themain di�erent form of alliances that exist and theirinformation needs while recognising, in addition,the phased development of trust already devel-oped. The results of this consideration are shownin Table 3.Tidd. et al. (1997) provided an analysis of types

of collaboration (that was adopted in slightlymodi®ed form in constructing Table 3) togetherwith evidence that speci®c forms of alliance tendto be linked with speci®c objectives.28 Table 3shows the types of collaboration in the ®rst col-umn with their associated purposes in the secondcolumn. The remaining columns then indicate thehypothesised information need according to thestage to which the relationship has developed.While this paper is devoted to identifying how

the provision of business information in generalneeds to be related to the building of trust rela-tionships in di�erent forms of associations, it ispossible to draw some conclusions relating speci-®cally to accounting from Table 3. There is clearlya di�erent emphasis in the detailed informationcontent required for di�erent types of alliance interms of information about technology, suppliersor pro®tability of separable business ventures.Nevertheless, the overwhelming characteristic ofthe information need is that, for the most part,existing accounting techniques seem to be ade-quate for addressing the di�erent speci®c issues ateach stage. There is, however, one particular fea-ture, over and above those identi®ed when dis-cussing business relationships earlier, wherefurther accounting research and practical devel-opment might usefully be focused. This area is nottotally new to the literature, but the previous workis not extensive and the analysis above suggeststhat more attention should be paid to it.The analysis emphasises the need, at the build-

ing commitment stage, to consider alternative

alliances that could be made. This indicates thatthere will be a need to place a value on the rela-tionship with each di�erent partner in order toassess the relative merits of alternate partneringrelationships. This can be linked directly to thegrowing body of literature, which addresses thequestion of how to make investments where cash¯ows are di�cult to assess,29 but little has yet beenwritten on placing a value on commercial associa-tions or relationships. Ford et al. (1998) addressedthis question and concluded that much mileagewas unlikely to be achieved by considering a DCFapproach to such a problem. They proposed thatone might attempt to ``value'' relationships usingschemes for weighting relationship attributes toproduce a form of utility index that could then beused instead of ®nancial return to conduct risk/return investment appraisals for competing asso-ciations. The weighting might also be linked to theperceived bene®ts that the would-be associatecould add to di�erent facets of the ®rm's valuechain.The analysis in Table 3 also emphasises that, once

a long term commitment has been established,recent developments like open book costing, a targetcosting approach to product development and thereturn/risk sharing modelling discussed earlier inthis paper come to the fore, especially in view ofthe increasing number of business alliances. Thisneeds to be tempered, however, with the earliercomments about changing accounting translucence.

7. Information needs and trust within networks

Having considered the nature of informationwithin relationships and the information needs ofdi�erent forms of alliances, the level of analysis canbe raised, at last, to the level of networks. As statedearlier, networks are more than bilateral alliances;they represent the con®guration of alliances, butalso both close and ``arm's length'' relationships inwhich the organisation is involved.

28 There are various ways of classifying types of business

alliances (viz. Tidd et al., 1997, p. 203, and their discussion of

horizontal/vertical; tactical/strategic classi®cations).

29 Areas of literature relating to investment in R&D, large-

scale computer systems and strategic investments all address

this type of problem.

182 C. Tomkins / Accounting, Organizations and Society 26 (2001) 161±191

To a large extent companies may not be able tomanage or shape the network within which theyreside. Large and powerful companies may bemore able to in¯uence its suppliers and customers,but even such companies will have to accept someaspects of its business network as given. Thismight relate to relationships with governments,®nancial institutions, universities, rival companies,etc. In any case, if an organisation wants tochange the nature of the network in which itoperates, it will have to do this through the med-ium of changing a series of bilateral alliances evenif there is a wider consensus for change.The majority of organisations will already be

located in a network. Hence, any e�ort to managethe network must normally be couched in terms ofnegotiated adjustments to something that existsrather than creation of something totally new.Nevertheless, to the extent that relationships canbe changed, some in¯uence over the network canbe exercised. So there is a managerial issue con-cerned with network design and this may beinterpreted as working through the followingquestions:

a. Is the organisation's information systemadequate for mastery of events?

b. Is the organisation's portfolio of alliancesand relationships (i.e. mix of trust intensities)appropriate for meeting its objectives?

Addressing question (a) ®rst, a network involvesmore than one alliance/relationship with high trustintensity and so the negotiation of roles, risks andpro®t-sharing will be more complex. Hence, themodelling of the type described in the Appendixwill become more complex. This is not only becausethere will be more parties to the negotiation, butalso because the simulations will need to captureboth the direct e�ects of decisions made by theorganisation and all its partners in bi-lateral rela-tionships and also the e�ects of decisions made byits partners with third or fourth parties in the net-work which may impact upon the ability of theorganisation to negotiate with any of those others.The next obvious question is whether the model-

ling will become too complex to function practically.At ®rst sight, the necessity to trace the impact ofchanges in bi-lateral relationships on all other

parties in the network seems likely to create con-siderable additional complexity. Complexity can,of course, be minimised by treating much of thenetwork (for example, transactions with ®rms withwhom one has just buying and selling transactionsin a fairly free market) as exogenous input to themodel. On the other hand, the increase in com-plexity may not be excessive.Tidd et al. (1997) say:

the number of alliances has increased overtime, and networks of collaboration appearto have become more stable, being basedaround a number of nodal ®rms in di�erentsectors (p. 215).

Partners of the nodal ®rm may come and go,but there is, it seems, the development of stablenetworks in most industries. Hence, the nodalcompanies might develop the models needed andeducate new partners accordingly. Movement inthis direction seems evident in the existence of leadcontractors in consortia operations such as thosebecoming more widespread in the UK since thedevelopment of the Conservative Government'sPrivate Finance Initiative or New Labour's Pub-lic±Private Partnership schemes. There is, though,also evidence that there is still much to be achievedto get the right balancing of risk and return insuch schemes which suggests that improved mod-elling is needed as the basis to negotiation (Wong,1999).One may also conclude that there is not so much

a need for new accounting techniques for themastery of events within networks, but rather aneed to consider how to use them in negotiationprocesses when there are more complex interac-tions across organisational boundaries.Question (a) concerns the process of organising

information exchange to master events given thestate of relationships in the network. Question (b)concerns whether the portfolio of relationships inthe network (i.e. the balance of trust intensitiesand information) is itself appropriate. To addressQuestion (b), ®rst, the organisation will need toreview the portfolio of relationships to decidewhether that mix of trust intensities is appropriatefor achieving its goals. This might be pursued by

C. Tomkins / Accounting, Organizations and Society 26 (2001) 161±191 183

mapping a speci®cation of the current network inwhich the organisation operates. Adapting Mitch-ell (1973), the four network dimensions wouldspecify for each of the bi-lateral alliances in thenetwork:

. the transactions needed,

. the degree of economic interdependence andvalue of the relationship,

. the current trust intensity and

. the current information (formal and informal).

This would incorporate relationships at di�erentstages of development and those ranging from``arm's length'' market transactions to close-knitassociations on di�erent forms of alliances. Tosupport the assessment of trust intensity there arealready some guidelines. Cummings and Bromiley(1996), for example, have developed and applied aconceptual basis and a series of questionnaires tomeasure organisational trust. Moreover, Landry(1998) has made a pioneering a attempt to relatetrust intensity to business e�ciency and concludedthat the most trusting US company in his studyachieved more than twice the dollar level of pur-chases per employee compared to the least trust-ing. There is certainly scope for improving themethodology for costing poor trust and the studiesjust mentioned are not comprehensive in addres-sing this, but they are a start and the process oftrying to build a map as suggested by Mitchellmight be very revealing for the organisation (orconsortium) concerned. It might even move ustowards providing a more rigorous theoreticalbasis upon which to determine cost-e�ectiveexchanges of business and accounting informationacross organisations.This process may reveal that trust intensities in

some relationships need to be changed, eitherbecause they are inconsistent with informationavailable or the degree of economic inter-dependence. This may lead to direct action to buildmore trusting relationships or even abandon themwhere new circumstances suggest that resorting toopen market operations might be bene®cial.Associated with developing or abandoning appro-priate trust intensities must come the appropriatebalancing of information.

At ®rst, this may seem too complex, but Tidd etal. (1997, p. 217) also show how di�erent indus-tries tend to specialise in di�erent forms of alliancein order develop innovative links for technologicalinnovation. If that holds across all purposes forwhich alliances are sought, speci®c ®rms willdevelop expertise for building relationships mostappropriate to its own sphere of operation and theproblem of assessing the portfolio of trust inten-sities will not be too severe. They can con®nethemselves to the relevant sections of Table 3.There is, however, one additional area of com-

plexity relating to trust intensity. The essence of aportfolio problem is that it is inappropriate toassess individual components on a separate basis.Hence, to judge whether the mix of relationships issatisfactory one cannot just assess whether theexisting trust intensities match what is requiredand try to change some of them, in isolation, ifnot. Changing trust in one alliance may a�ecttrust in other areas of the network. For example,is there an organisational parallel to personal jea-lousy- perhaps through members of the outer net-work being suspicious of the motives of an innercore membership or because relationships betweenthe new closer partner and others in the networkare already poor? Is there a limited capacity togive close attention to more than one or two rela-tionships? If answers to these questions are a�r-mative, they suggest that trust in networks has anextra dimension to that in isolated bi-lateral alli-ances. Part of the balancing of trust and informa-tion may have to consider the ``three's a crowdphenomenon''.

8. Summary of arguments and their signi®cancefor research and practice

The initial motivation for this paper was toexplore the likely in¯uence of the growth in busi-ness networking on business information. At the®rst level at which this was addressed, it seemedclear that there will be a greater emphasis on theuse of management accounting within inter-orga-nisational negotiation processes and it was pro-posed that this would mean that accountants

184 C. Tomkins / Accounting, Organizations and Society 26 (2001) 161±191

would have to be more agile in providinginformation needed to present and respond toarguments. This is likely to lead to more complexcomputer simulations and it must be recognisedthat while relevant information in¯uences nego-tiation processes, it is also a function of them. Therole and signi®cance of both accountants andaccounting in alliances cannot be fully assessedwithout an understanding of speci®c negotiationprocesses.The analysis was then pursued at a deeper level.

It was felt to be necessary to examine in detail howinformation is needed to support ®rst relationshipsand then alliances, both of which are the basicbuilding blocks of networks. This called for aconsideration of the concept of trust and howtrust intensity a�ects information provision. If, asin this paper, trust is de®ned to mean areas of lifewhich one can take as given, trust must be a funda-mental factor in deciding what amount and type ofinformation should be presented. The considerationof trust is the fundamental consideration before allother determinants come into play. Trust asde®ned here will determine where the informationsystem needs to focus and with what intensity. It issurprising, therefore, that this dimension seems tobe absent from accounting theory, although it isobviously implicit in many accounting procedures.Quite apart from the debate about alliances, thispaper argues for more work to develop explicittheories of how trust needs to be taken into accountin all the di�erent dimensions of accounting. Sucha step has promise for laying a more rigorous basisfor accounting theory and practice.A conceptual examination of the relationship

between trust and information suggested that theconventional argument that there is a simplemonotonic inverse relationship between employ-ment of trust and the provision of information issimplistic. This only holds at a speci®c point oftime for a speci®c decision. An inverse U-shapedfunction is appropriate when the dynamics ofrelationship life cycles are considered . It was alsofound to be conceptually useful to distinguishbetween information needed to create and supportcontinuing trust (Information Type 1) and infor-mation needed for mastery of events (InformationType 2). Both types exhibit, over time, an inverse

U-shape function between trust intensity andinformation. This further suggests that a model ofinformation redundancy could be constructedbased upon the concept of trust. An under-standing of what this entails might enableaccountants to keep their systems lean, up-to-dateand more cost-e�ective. The practical implicationsof this could be very far reaching.While it is useful to make the conceptual dis-

tinction between the two types of information,Information Type 1 and Information Type 2 arenot totally separate from each other. InformationType 1 will directly a�ect the degree of trust andthen subsequently be in¯uenced by it. Trust inten-sity will also show what cannot be taken as givenand in¯uence the focus of control behaviour,which, in turn, generates the demand for Infor-mation Type 2. Interaction over Information type2 will also provide experience that will in¯uencethe state of trust and, through that, feed backinto demands for more or less Information Type1. These are, however, propositions deductivelyderived. Empirical research is needed test thesepropositions and this would entail unravellingthe network of direct and indirect in¯uencesbetween Information Types 1 and 2 and trustintensity.The paper also explored whether propositions

based on interpersonal trust can be applied tointer-organisational trust. It was concluded that,with modi®cations, they can. Moreover, the exis-tence of contracts as the basis of alliances does notmake the argument over the trust/information mixirrelevant. In fact this mix should be addressed informing inter-organisational contracts. Researchon the extent to which contracts already prescribeinformation exchange would be valuable. Eventhen, it is doubted that contracts can completelyprescribe the desirable trust/information mix andso the issue needs regular attention by controllersin alliances.No new basic accounting techniques seem to be

required to cope with operation in alliances andnetworks, but the need to trace the impact ofplanned and actual events across two organisa-tions will need more complex analysis in mostareas of accounting. Recent developments likeopen book costing and target costing seem par-

C. Tomkins / Accounting, Organizations and Society 26 (2001) 161±191 185

ticularly relevant. However, it was argued thataccounting techniques should not be introducedjust because they seem rationally apposite. Theirimpact on trust in the relationship should alsobe considered and the notion of opaqueness,translucence and transparency in informationsystems o�ers a valuable area of enquiry both inrelation to alliances and in the wider ®elds ofaccounting.It is clear that di�erent forms of business alli-

ances have di�erent information needs, but all ofthem exhibit a staged development of the balancebetween trust and information as the alliancedevelops and further investigation is needed toestablish in more detail how accounting informa-tion ¯ows need to di�er according to the type ofalliance.Moving ®nally to multi-party networks, more

complex simulation models will be needed to sup-port negotiation over sharing tasks, risks andpro®ts in order to master events, but they may notneed to be excessively complex. Networks bringwith them, however, a new problem of establish-ing an appropriate portfolio of trust intensitiesand associated information requirements. Organi-sations might gain bene®t from mapping networksalong four dimensions: transactions needed, eco-nomic interdependence, trust intensity, informa-tion needed. This line of thinking may also lead toa more rigorous theory for cost-e�ective informa-tion system design.Quite far-reaching implications have been

drawn in bringing in a consideration of trust intothe design of information systems. One mustremember, nevertheless, that, while some elementof trust is essential for life to proceed, custom andpractice can change the level of trust intensityemployed. One may question whether the recentgrowth in business alliances and networking willbe a persistent phenomenon on such a widespreadbasis. The growth of alliances has followed largelyin the footsteps of Japanese and Continental Eur-opean success in utilising close business relation-ships. Furthermore, Japanese alliances dependsigni®cantly upon the power of nodal ®rms andthe network in general to exercise sanctions fortrust breaking (Hagen & Choe, 1998). Will net-works and nodal ®rms be able to maintain such a

grasp of events in a world ever more open tointernational competition? More recently generalopinion seems to be that the greater American(and to a lesser extent UK) reliance on marketswill return to dominate. The recent (1999) recov-ery in the Japanese and German stock marketswas largely predicated upon the assumption thatthis will happen. Moreover, will a di�erent formof network, the Internet, take over from closepersonal relationships and, if so, can trust be builtup to the same extent when interactions are elec-tronic-based rather than personal? Will E-com-merce facilitate the development of a more globalmarket whereby it will be easier and more eco-nomic to operate as infrequent buyers and sellersrather than close associates? To some extent theneed for close and trusting relationships may les-sen and save us from the more complex modelssuggested in this paper.The Internet will, however, also aid the devel-

opment of collaboration. Many examples alreadyexist of companies exchanging information, in realtime, relating to procurement, joint productdevelopment and investment. In principle, thisshould involve no di�erent concepts from thosedescribed in this paper, but it will need consider-able application from accountants and informa-tion managers. The inter-organisation costbehaviour e�ects will need to be readily seen atlower levels of the organisation if errors are to beavoided, perhaps on a free access basis. On theother hand, the development of capable ``®re-walls'' and translucent systems to prevent the lossof key information too early in the relationshipwhen costs of betrayal are low will also needattention. These current developments in businessorganisation and information must be recognisedand taken into account in attempts to testempirically the propositions expressed in thispaper.Collaboration both between species and within

species is evident throughout the natural worldwhere competition to survive is severe and energyis not expended needlessly. I see no reason why thesame should not apply to the business world.Irrespective of the current state of technology,alliances will be pursued and trust given wherethat approach to reducing uncertainty supports

186 C. Tomkins / Accounting, Organizations and Society 26 (2001) 161±191

mutual survival. Writing about the natural world,Jones (1999, p. 44) states:

Without collective trust, it (science) could notwork. Instead there would be the dismalapparatus of mutual suspicion familiar toevery accountant.

It is time for accountants to demonstrate thatthey do not distrust everyone and, hence, do nottry to insist on excessively detailed information. Itis time for them to make their systems lean.Recent discussions of lean accounting (e.g. Mas-kell, 1996), have emphasised more the need to besparse with supplying information Type 2. Evenmore leanness might be obtained by balancingtrust and information more carefully.

Acknowledgements

This paper results from the author being invitedto attend The University of Uppsala Workshop onBusiness Networks and Accounting, Sigtuna,Sweden, September, 1999. Although now sub-stantially extended, this paper would not havebeen written had there not been an invitation topresent thoughts on this topic. Thanks are alsodue to Anthony Hopwood for his encouragementto extend the paper and o�er it for critical debateand also to Louise Young, Robin Holt and twoanonymous, but helpful, reviewers for their com-ments.

Appendix A. Illustrating the use of simulation toprovide mastery of events for each partner and anacceptance of fairness.

Table A1 is based on the assumption that two®rms, A and B, are to collaborate over an invest-ment project and that A is to provide some tangi-ble goods as inputs to B's processes.30 Let usassume that, at ®rst, each controller says initially

that the relationship seems to be fair if each earnsthe same IRR on their own investment as thatearned by the project as a whole. It is possible toset up a simple simulation on variations in thetransfer price to achieve that. Table A1 showsthat, for the cash ¯ows assumed, the project31

earns an IRR of 27% and each ®rm will achievethat IRR at a transfer price of £1588.One controller may then argue that, while this

yields the same expected return for both parties, itdoes not address the extent to which the two ®rmsface di�erent risks. Hence, A and B decide toanalyse relative risks in detail with a view toadjusting the transfer price, and hence give a lar-ger expected return, to compensate the partybearing most risk. As a means of doing this aMonte Carlo Analysis32 has been run on the cash¯ows for this project to develop the full prob-ability distribution of IRRs earned by (a) theproject as a whole and (b) each ®rm at a transferprice of £1588. Various underlying probabilitydistributions have been assumed for the di�erent®gures in Table A1 and put into the spreadsheetusing the @Risk software.33 The results of this®rst simulation are shown in Table A2 under thelabel Simulation 1.

30 This example is framed as a transfer-pricing situation, but

similar examples could be constructed for any form of business

association.

31 The project illustrated is, of course, an extremely simple one,

but this Appendix is concerned just to demonstrate the type of

interaction between the parties to a relationship and how this is

subject to and developed by a series of calculations (simulations).32 In fact, a more e�cient variation on Monte Carlo sam-

pling has been used called Latin Hypercube sampling using the

@Risk package, but it should be borne in mind that this form

of simulation has only been used to illustrate general concepts.

These inter-party negotiations could be conducted through the

medium of any type of calculation that was thought to be

appropriate. This Appendix is not meant to be advocating the

use of @Risk nor, indeed, any particular type of simulation Ð

probabilistic or otherwise.33 The @Risk software permits a user to employ a high

variety of forms of probability distributions. This simple

example uses normal distributions for all variables, but the

principles apply equally whatever form or mix of probability

distributions is used in the model. The distributions have been

assumed to be independent of each other in this example, but,

in practice, it would be necessary, and quite easy, to allow for

interdependence between the uncertainty attached to di�erent

variables in the model.

C. Tomkins / Accounting, Organizations and Society 26 (2001) 161±191 187

Recognising the risks involved shows that B ismost unlikely to be satis®ed with the ®nancialarrangement speci®ed in Table A1. It becomesquite clear that, although B has the same expectedreturn of 27%,34 it is subject to much more riskthan A. It has a far larger downside risk and asomewhat lower probability of achieving even apositive rate of return35.

Let it be assumed that B, once this is recognised,wants a lower transfer price to compensate for theextra risk it faces and proposes that the transferprice should probably be as low as £1,400. Ofcourse, the project IRR will not change, but themanual simulation showed that A will now beexpected to earn an IRR of just 12% whereas Bwould now get 30%. Now A will want to look at

Table A1

Decomposition of project cash ¯ows to give parties an equal IRRa

a. Total project appraisal

Year A cash

outlays (£)

B cash

outlays (£)

Volume

sold

Final

product price (£)

Revenue (£) Net cash

¯ows (£)

0 100,000 600,000 ÿ700,0001 40,000 150,000 60 10,000 600,000 410,000

2 35,000 150,000 80 10,000 800,000 615,000

Project

IRR 27%

b. Decomposition into the separate returns to each ®rm

Simulation of the transfer price B pays to A £1588

Firm A Firm B

Year Cash

outlays

Transfer

price

Volume Net

cash ¯ow

Cash

outlays

Final

product price

Net cash

¯ows

£ £ £ £ £ £

0 100,000 ÿ100,000 600,000 ÿ600,0001 40,000 1588 60 55,280 150,000 10,000 354,720

2 35,000 1588 80 92,040 150,000 10,000 522,960

IRR(A) 27% IRR(B) 27%

a The use of IRR is just for simplicity. One could use CAPM based discounting or any other measure of pro®tability instead of or in addition to IRR.

34 It can be seen from Table A2 that the Monte Carlo

simulation generates slightly di�erent expected rates of return

from the 27% generated in the manual simulation in Table A1.

The use of more iterations in the Monte Carlo simulation,

would gradually reduce and, in theory, eventually remove these

di�erences. There is an important practical lesson here. The

author used 10,000 iterations to generate the results in Table

A2 and, even with this very simple model, the results still did

not stabilise on the IRRs calculated on the expected values of

each variable. Table A2 indicates simulated IRRs that are

di�erent by several per cent from the appropriate outcome.

Practising managers must not simply assume that, say, 2000

iterations are su�cient and accept the results blindly. This

would become more problematic in real life investment

problems. The author has experimented with more complex

models and it can take 30,000 plus iterations to achieve

stability. Even with modern PCs, this can take several

hours and, hence, limits the value of this type of simu-

lation actually within the negotiations on a real time basis.

The di�erences between the IRRs generated in the manual

and Monte Carlo simulations have been shown in Table A2

to warn of this practical issue. For the purposes of the dis-

cussion in this section of the paper, attention should be

focused on the risk descriptors being aware that they also

might be slightly di�erent if many more iterations were

employed.35 The @Risk package provides much more detail on the

distributions than shown here. One can also choose to work

with various measures of risk including variance. The measures

shown in Table A2 are just illustrative of risk measures that can

be generated. They may, of course, be far simpler and not

based upon probabilistic data at all.

188 C. Tomkins / Accounting, Organizations and Society 26 (2001) 161±191

the risk attached to the 12% return. B will knowthat its risk has been reduced, but it will also wantto see by how much.The revised position on relative risks is shown

by Simulation 2. It can be seen that A would nowbe expected to earn 12%, but now has only an85% chance of a positive return and the maximumdownside increases to ÿ33%. B achieves a smallincrease in expected return, but does not achievemuch risk reduction at all. Consequently, A mightargue (assuming they are operating the modelopenly together) that there seems little point inmaking its position signi®cantly worse when Bdoes not bene®t greatly Ð this occurs because B'sinvolvement in the project is more extensive. So

suppose that A now suggests that, in order toreduce B's risk, A pays B a ®xed sum of £100,000at the outset of the project as its contribution toinitial development costs incurred by B and thatthis payment is incorporated in a revised manualsimulation to ®nd the transfer price which thengives both A and B an expected IRR equal to theproject 27% IRR. Hence, the terms of the agree-ment between A and B would include a ®xedinitial payment from A of £100,000 and the trans-fer price used would be as revised. Under thissuggestion the revised manual simulation showedthat both parties would be expected to achievea 27% IRR at a transfer price of £2626. Theassociated revised risk pro®les are shown in

Table A2

Monte Carlo simulation to support negotiation in alliancesa

Project Firm A Firm B

Simulation 1 TP=1588

Mean IRR from simulation (%) (%) (%)

31 28 29

Maximum 302 79 326

Minimum ÿ310 ÿ25 ÿ401probability that IRR is positive 84 98 80

Simulation 2 TP=1400

IRR on expected values of each cash ¯ow 27 12 30

Mean IRR from simulation

31 12 31

Maximum 271 54 310

Minimum ÿ373 ÿ33 ÿ376Probability that IRR is positive 84 85 82

Simulation 3 TP=2626 A pays B £100,000 ®xed payment for research

IRR on expected values of each cash ¯ow 27 27 27

Mean IRR from simulation 32 27 23

Maximum 299 79 422

Minimum ÿ372 ÿ207 ÿ672Probability that IRR is positive 84 99 73

Simulation 4 TP=2626 Research

contracted out £420,000 of

which A pays

for ®xed payment

of £150,000

IRR on expected values of each cash ¯ow 25 10 33

Mean IRR from simulation 25 10 34

Maximum 65 44 159

Minimum ÿ32 ÿ26 ÿ411+probability that IRR is positive 96 87 96

a These maximum and minimum returns look extreme and are due to the shape of the tails of a Normal distribution. More realistic values would be

achieved by using distribution forms with truncated tails. There are several of these in the @Risk package including discrete, triangular and TNormal

distributions. This does not a�ect, however, the principles being discussed in this Appendix.

C. Tomkins / Accounting, Organizations and Society 26 (2001) 161±191 189

Simulation 3 in Table A2. The results show thatA, rather than B, bene®ts from risk reduction. A isnow 99% likely to avoid loss, although there is avery small probability of a large downside risk ofup to ÿ207%. B's risks increase signi®cantly interms of downside potential and probability ofpositive returns. So this ®nancial arrangementdoes not seem to be the answer.Now suppose that the two parties start to feel

uncomfortable with the whole project risk andagree to engage in risk management to reduce thestotal risk that has to be allocated betweenthem. B then analyses its initial capital outlaysmore cloely and notes that the probability dis-tribution entered into the Monte Carlo simula-tion in the form @NORMAL(600000,250000)36

to indicate considerable uncertainty could besplit as follows:

Fundamentalresearch

@NORMAL (400000,200000)

Development andcapital equipment

@NORMAL (200000,50000)

and that the fundamental research would be con-tracted out to a third party specialist for a ®xedpayment of £420,000 of which A would pay£150,000 and B the balance. The transfer price of£2626 would be retained. The manual simulationshowed that this arrangement would yield a pro-ject IRR of only slightly less at 25% and A wouldget 10 and B 33%. The revised risk positions areshown in Table A2 under Simulation 4. Now A'sposition is not a great deal di�erent from Simula-tion 2, but at least B's position seems improved.It matters not whether A and B would settle on

this arrangement. They might continue to tryalternative ®nancial arrangements until a ®nancialbasis for fair collaboration was found. The pointhas simply been to illustrate how simulation (i.e.repeated revisions of calculation and information)will be needed to help shape the project anddevelop mutual understanding of what itinvolves.

References

Aulakh, P., Kotabe, M., & Sahay, A. (1997). Trust and per-

formance in cross-border marketing partnerships. In P.

Beamish, & J. Killing, Cooperative strategies: Vol 1. North

American perspectives (pp. 163±196), New Lexington Press.

Axelrod, R. (1984). The evolution of co-operation. New York:

Basic Books.

Barber, P., Graves, A., Hall, M., Sheath, D., & Tomkins, C.

(2000). Quality failure costs in civil engineering projects.

International Journal of Quality and Reliability Management,

17(4/5), 479±492.

Beale, H., & Dugdale, A. (1975). Contracts between business-

men: planning and the use of contractual remedies. British

Journal of Law and Society, 2, 45±60.

Brennan, R., & Turnbull, P. (1998). Adaptations in buyer-seller

relationships. In P. Naude , & P. Turnbull, Network dynamics

in industrial marketing. Pergamon.

Carr, C., Tomkins, C., & Bayliss, B. (1994). Strategic invest-

ment decisions: A comparison of UK and German practices in

the motor components industry. ICAE&W/Avebury.

Castells, M. (1996). The rise of the network society. Blackwell.

Creed, W., & Miles, R. (1996). Trust in organizations: a con-

ceptual framework linking organizational forms, managerial

philosophies and the opportunity costs of controls. In

Kramer, R. & Tyler, T., Trust in organizations: Frontiers of

theory and research (pp. 16±38), Sage.

Cummings, L., & Bromiley, P. (1996). The organizational trust

inventory (OTI). In Kramer, R. & Tyler, T., Trust in organi-

zations: Frontiers of theory and research (pp. 302±330). Sage.

Das, T., & Teng, B. (1998). Between trust and control: devel-

oping con®dence in partner cooperation in alliances. The

Academy of Management Review, 23(3), 491±512.

Das, T., & Teng, B. (in press). Trust, control and risk in strategic

alliances: an integrated framework. Organisation Studies.

Debresson, C., & Amesse, F. (1991). Networks of innovators: a

review and introduction to the issues. Research Policy, 20,

363±379.

Faulkner, D. (1999, November). Trust and control in strategic

alliances. Financial Times: Mastering Strategy Supplement

(p. 29).

Feltham, G. (1968, October). The value of information. The

Accounting Review, (pp. 684±696).

Feltham, G., & Demski, J. (1969, July). The use of models in

information evaluation. The Accounting Review, (pp. 457±466).

Ford, D., McDowell, R., & Tomkins, C. (1998). Exploring

relationship strategy. In P. Naude , & P. Turnbull, Network

dynamics in international marketing. Pergamon.

Ford, D., Gadde, L., HaÊ kansson, H., Lundgren, A., Snehota,

I., Turnbull, P., & Wilson, D. (1998). Managing business

relationships. John Wiley and Sons.

Frank, R. (1988). Passions with reason: The strategic role of

emotions. New York: Norton.

Ghoshal, S., & Moran, P. (1996). Bad for practice: a critique of

transaction cost theory. Academy of Management Review, 21,

13±47.

36 A Normal distribution prescribed by a mean of £600,000

and a standard deviation of £250,000.

190 C. Tomkins / Accounting, Organizations and Society 26 (2001) 161±191

Gioia, D. (1986). Symbols, scripts and science making. In Sims,

H., Gioia, D. et al., The thinking organisation. Jossey-Bass.

Goldratt, E. (1990). Theory of constraints. Croton-on-Hudson,

NY: North River Press.

Hagen, J., & Choe, S. (1998). Trust in Japanese inter®rm rela-

tions: institutional sanctions matter. Academy of Manage-

ment Review, 23(3), 589±600.

HaÊ kansson, H., (1982). International marketing and purchasing

of industrial goods. John Wiley and Sons.

Harrison, R., Dibben, M., & Mason, C. (1997). The role of

trust in the informal investor's investment decision: an

exploratory analysis, Entrepreneurship Theory and Practice,

Summer, 63-81.

IMEDE. (1986). Ford of Europe (A), case study GM-321.

Jones, S. (1999). Almost like a whale: The origin of the species

updated. Doubleday.

Jones, T. (1995). Instrumental stakeholder theory: a synthesis

of ethics and economics. Academy of Management Review,

20, 404±437.

Lamming, R., Caldwell, N., & Harrison, D. (1999). The con-

ceptual background to transparency in supply, Paper pre-

sented at the British Academy of Management, Annual

Conference, The Manchester Metropolitan University, 1±3

September.

Landry, J. (1998). The value of trust. Harvard Business Review,

76(1), 18±19.

Lewicki, R., & Bunker, B. (1996). Developing and maintaining

trust in work relationships. In R. Kramer, & T. Tyler, Trust

in organizations: Frontiers of theory and research, (pp. 114-

139) Sage.

Luhmann, N. (1979). Trust and power, John Wiley and Sons.

Macaulay, S. (1963). Non-contractual relations in business: a

preliminary study. American Sociological Review, 55, 55±69.

Macaulay, S. (1977). Elegant models, empirical pictures and the

complexities of contract, Law and Society, Winter, 507±529.

Macaulay, S. (1985). An empirical view of contract, Wisconsin

Law Review, 465±482.

Madhok, A. (1995). Revisiting multinational ®rms' tolerance

for joint ventures: a trust-based approach. Journal of Inter-

national Business Studies, 26, 117±137.

Maskell, B. (1996). Making the numbers count: The accountant

as change agent on the world class team. Portland, OR: Pro-

ductivity Press.

McKnight, D., Cummings, L., & Chervany, N. (1998). Initial

trust formation in new organisational relationships. Academy

of Management Review, 23(3), 473±490.

Meyerson, D., Weick, K., & Kramer, R. (1996). Swift trust and

temporary systems. In Kramer, R., & Tyler, T., Trust in

organizations: Frontiers of theory and research (pp. 166±195).

Sage.

Mitchell, J. (1973). Networks, norms and institutions. In Bois-

sevain, J. & Mitchell, J., Network analysis: Studies in human

interaction. Monton and Co.

Monden, Y. (1989). Full cost based transfer pricing in the

Japanese auto industry: risk sharing and risk spreading

behaviour. In Monden, Y. & Sakurai, M., Japanese manage-

ment accounting. Productivity Press.

Nicolini, D., Tomkins, C., Holti, R., Oldman, A., & Smalley,

M. (in press), Can target costing and whole life costing be

applied in the construction industry? Ð evidence from two

case studies. British Journal of Management.

Power, M. (1994). The audit explosion. Demos: London.

Power, M. (1996). Making things auditable. Accounting

Organizations and Society, 21, 289±314.

Power, M. (1997). The audit society: Rituals of veri®cation,

Oxford University Press.

Reina, S., & Reina, M. (1999). Trust and betrayal in the work-

place. San Franscisco: Berrett-Koehler.

Rhys Jones, S. (1991). The in¯uence of law, language and per-

ception on con¯ict in the construction industry. MSc dis-

sertation, Bartlett School of Architecture.

Ring, P., & Van de Ven, A. (1994). Developmental processes of

cooperative interorganizational relationships. Academy of

Management Review, 19, 90±118.

Rosenberg, D., & Tomkins, C. (1983, September/October). The

budget liaison o�cer Ð guardian or advocate. Local Gov-

ernment Studies (pp. 51-64).

Quinn, R., & Spreitzer, G. (1997). The road to empowerment:

seven questions every leader should consider. Organisaza-

tional Studies, 26(3), 37±49.

Tidd, J., Bessant, J., & Pavitt, K. (1997). Managing innovation.

John Wiley & Sons.

Tomkins, C. (1991). Corporate resource allocation: Financial

strategic and organisational perspectives. Blackwell.

Tomkins, C. (1999). Using the theory of constraints and cost of

poor quality calculations to improve business processes. In

Oldman, A. & Mills, R., Handbook of cost management.

CIMA/Gee Publishing.

Tomkins, C., & McAulay, L. (1992). Modelling fair transfer

prices where no market guidelines exist. In Drury, C., Man-

agerial accounting handbook. CIMA.

Urmson, J., (1988). Aristotle's ethics. Blackwell.

Vincent-Jones, P. (1989). Contract and business transactions: a

socio-legal analysis. Journal of Law and Society, 16(2), 166±186.

Wicks, A., Berman, S., & Jones, T. (1999). The structure of

optimal trust: moral and strategic implications. The Academy

of Management Review, 24(1).

Wildavsky, A. (1964). The politics of the budgetary process.

Little Brown.

Wong, M. (1999). Risk assessment and risk allocation in IS/IT

private ®nance initiatives. Unpublished PhD thesis, School of

Management, University of Bath.

Zaheer, A., & Venkatraman, N. (1995). Relational governance

as an inter-organizational strategy: an empirical test of the

role of trust in economic exchange. Strategic Management

Journal, 16, 373±392.

C. Tomkins / Accounting, Organizations and Society 26 (2001) 161±191 191