Framing shared services

Critical Perspectives on Accounting xxx (2014) xxx–xxx

G Model

YCPAC-1824; No. of Pages 14

Contents lists available at ScienceDirect

Critical Perspectives on Accounting

journal homepage: www.elsev ier .com/ locate /cpa

Framing shared services
Accounting, control and overflows
Gustaf Kastberg

Department of Service Management, Lund University, Sweden

A R T I C L E I N F O

Article history:

Received 20 June 2012

Received in revised form 22 January 2014

Accepted 24 January 2014

Available online xxx

Keywords:

Management control

Public sector

Framing

Overflow

A B S T R A C T

During the last decade there has been an increase in the use of horizontal organizational

relationships (HORs) such as strategic alliances, shared services, and outsourcing. Studies

accentuate the unstable, fragile nature of HORs and the importance of directing attention

to how these arrangements are formed and how and why they change. The aim in this

study is to explore control problems of HORs and to discuss how different problems are

related to accounting and control initiatives. A theoretical framework with the core

concepts of framing and overflow is used as a theoretical point of departure. The empirical

study is a longitudinal case study of the organization of a shared service center. In this

study, we observe the problems of risk and coordination often discussed in the literature.

However, guided by our theoretical framework, we also observe how many problems were

formative and related to the forming of actors, relations and products. This study adds to

the literature by conceptualizing how accounting and control not only functioned as

framing devices, but also became sources of overflows, causing destabilization. Following

the conceptual framework, we observe how overflow was caused both through

reconnecting what had previously been separated and through cutting off relations in

the process of disentangling the HOR.

� 2014 Elsevier Ltd. All rights reserved.

1. Introduction

During the last decade there has been an increase in the reliance on cooperative inter- and intra-organizationalarrangements such as strategic alliances, shared services, and outsourcing (Broadbent and Guthrie, 2008; Cristofoli et al.,2010). The trend has also been observed in the public sector where more organizational solutions presuppose sharperboundaries between departments and where competition often arises between units in the same, overarching organization(Christensen and Lægreid, 2007; Peters, 1998). Horizontal organizational relationships (HORs) create new challenges fororganizers and have been characterized as a hybrid (Kurunmaki and Miller, 2011) since they are neither purely verticalorganizational arrangements nor purely market arrangements. The private sector literature also indicates that suchhorizontal cooperative arrangements frequently fail (Meira et al., 2010) and the public sector literature indicates that HORsare difficult to coordinate (Christensen and Lægreid, 2008).

Several studies indicate that it is important to consider the complexity of the situation and the dynamics of the HORs asthey evolve over time. Studies have, for example, shown that control initiatives might gain or lose relevance over time (Chuaand Mahama, 2007; Thrane, 2007), that interacting parties change their identity vis-a-vis the relations (Mouritsen et al.,2001), that inter-organizational control affects the intra-organizational control (Carlsson-Wall et al., 2011; Thrane and Hald,

E-mail address: [email protected].

Please cite this article in press as: Kastberg G. Framing shared services. Crit Perspect Account (2014), http://dx.doi.org/10.1016/j.cpa.2014.01.002

1045-2354/$ – see front matter � 2014 Elsevier Ltd. All rights reserved.

http://dx.doi.org/10.1016/j.cpa.2014.01.002


mailto:[email protected]



http://www.sciencedirect.com/science/journal/10452354

www.elsevier.com/locate/cpa


G. Kastberg / Critical Perspectives on Accounting xxx (2014) xxx–xxx2

G Model


2006) and that a lot of energy is spent on organizing the relations (Janssen and Joha, 2008; Lindvall and Iveroth, 2011). Alinear perspective on the development of relations and the use of control has also been questioned. Studies also indicate theneed to understand that there are several logics and centers that influence the situation (Thrane, 2007).

These studies accentuate the unstable, fragile nature of HOR and the importance of directing attention to how thesearrangements are formed and how and why they change (Chua and Mahama, 2007; Mouritsen et al., 2001; Thrane, 2007;Thrane and Hald, 2006). Following Callon’s (1998b) reasoning about framing and reframing of relations, we can expect thataccounting and control initiatives might be a source contributing to stabilization and preservation of existing states, but alsoa source of destabilization. While the former have been observed and discussed more often in the literature (Dekker, 2004;Mouritsen et al., 2001; Tomkins, 2001; Vosselman and van der Meer-Kooistra, 2009), the latter has less often been discussed.However, following the reasoning of Callon, accounting is a practice which, alongside other metrical activities, is necessary inorder to bring about a problematization of existing frames.

In order to contribute to the growing literature in this area, this study explicitly addresses a central issue: the underlyingproblems of HORs and the role accounting and control play. The question might seem trivial, but in their comprehensivereview, Caglio and Ditillo (2008) conclude that while there is quite an extensive body of literature on control (that is,solutions), there are few studies that actually focus on the problems that create the need for control. When addressing theunderlying problems, both the sources of stabilization and change become part of the analysis (Beckert, 2009; Caglio andDitillo, 2008; Thrane and Hald, 2006). As we change focus from control solutions to control problems, there is an opportunityto scrutinize relations between perceived problems and control measures taken. Consequently, the aim of this study is to

explore control problems of horizontal organizational relations and to discuss how different problems relate to accounting and

control initiatives. The research question addressed is therefore how accounting and control contribute to both thestabilization and the destabilization of HORs.

The theoretical framework used is Callon’s (1998a, 1998b) theorizing about framing and overflow and the empirical studypresented is a longitudinal study of the organization of a shared service center. In this study, we observe the problems of riskand coordination often discussed in the literature. However, we also observe how many problems were formative andrelated to the forming of actors, relations and products. We observe how accounting and control contributed to thestabilization of the HOR, which is in line with observations made by others (Coad and Cullen, 2006; Kurunmaki and Miller,2011; Mouritsen et al., 2001). This study adds by also conceptualizing how accounting and control not only functioned asframing devices, but also became sources of overflows, causing destabilization. Following Callon’s conceptual framework, weobserve how overflow was caused both through reconnecting what had previously been separated and through cutting offrelations in the process of disentangling the HOR.

The paper is arranged as follows: first, a theoretical framework based on Callon’s (1998b) concepts of framing andoverflow is presented. A theoretical discussion follows, which has implications for the analysis and the empiricaldescriptions. Thereafter, a longitudinal case study of the organization around a shared service center is presented. In theanalysis, the framing process is described with a focus on the problems that emerged and how the parties tried to handlethem. Next, a discussion is developed about sources of overflows. The paper concludes with some general conclusions drawnfrom the research.

2. Framing and overflow – a theoretical framework

Callon’s (1998a) concepts of framing and overflow are at the center of this study’s theoretical framework. As indicated inthe introduction, one important argument that will be discussed is that we might expect accounting and control not only tobe a source of stabilization for the interacting parties but also a source of destabilization. The theoretical frameworkpresented helps connect accounting to the entities counted and the organizational actors that use, shape and are shaped byaccounting.

The concepts of framing and overflow are rooted in actor-network theory (ANT). However, when introducing theconcepts, Callon (1998b) explicitly referred to Goffman (1986). In their review of the use of ANT in accounting studies, Baxterand Chua (2003) concluded that there is an established tradition in accounting that departs from ANT, although it can hardlybe described as a mainstream approach. Most scholars still refer to the early work of Latour (Mouritsen and Justesen, 2011),while there are some who refer to Callon’s framing and overflow framework (Kastberg and Siverbo, 2008; Lowe, 2001).Others have noted that there is a tendency to focus on the constructivist dimension in the analysis – on how structures arestabilized – with limited focus on the causes and effects of established networks and structures (Lowe, 2001). The use of theconcepts of framing and overflow directs attention to the establishment of entities and relations, to the effects caused by theframing process, and not least, to the fragility of the relations (Barry and Slater, 2002; Callon, 1998b).

2.1. Framing

Callon (1998b) addresses the question of how economic interaction is possible; the answer being that a frame in whichinteraction can take place is established. This frame stabilizes the conditions and makes it possible to calculate the outcomeof the interactions. Only when possible to count on, is economic interaction made possible. The point of departure, whenusing Callon’s framework for theorizing about actors and interactions, is that neither the actors nor the products are to betaken for granted as ready-made and stable entities. One of the main arguments is that because economic actors and their




G. Kastberg / Critical Perspectives on Accounting xxx (2014) xxx–xxx 3

G Model


forms of interactions are created, the maintenance of the stabilized world that allows interaction needs ‘‘substantialinvestments’’ (Callon, 1998b).

While economists make the mistake of viewing ‘‘economic man’’ as given by nature, sociologists tend to make the mistakeof declaring that economic man is an impossibility because of his social nature (Granovetter, 1985). What Callon emphasizesis that economic man is neither natural nor impossible. Instead, economic man is a role that is played in certaincircumstances; in a market context. An advantage with stabilized identities in market oriented relationships is that theinteracting parties are aware of each other’s underlying motives. Several scholars have criticized the mainstream HORliterature for being reductionist in its view of actors involved in the relations (Chua and Mahama, 2007). When actors aretreated as black boxes, the interaction between the inner processes of the actor and the relations has been neglected. Both theactor and the interactions must, in fact, as Granovetter (1985) argues, be understood in relation to each other. That actors inHORs are affected and transformed in relation to the interactions is well illustrated in a study by Mouritsen et al. (2001). Thereasoning indicates the importance of considering the framing of the identities of the interacting parties in relation to theHORs.

The literature gives several other examples, besides the framing of identities, of dimensions in the framing process thatare important to consider. Contracts and coordination of exchange constitute core activities in framing exchange relationsthat have been described (Callon, 1998b; Caglio and Ditillo, 2008). One central dimension of importance in the framingprocess is the ‘‘object’’ that the relations center around (Beckert, 2009; Callon, 1998a; Slater, 2002). Sometimes, when thereis an easily defined product at the center of the cooperation, the parties can effortlessly agree on the product’s relevantproperties. This is, for example, the case when the relations center on standardized, physical products. Often, and not leastwhen organizing IT services as in the case presented in this paper, HORs are created around something less specific wheredifferent dimensions might be contested and the definitions are not easily agreed upon (Beckert, 2009; Callon et al., 2002;Orlikowski and Iacono, 2001; Quattrone and Hopper, 2006).

Callon et al. (2002) refer to the ‘‘economy of qualities’’ when addressing the service economy. The reason is the specificchallenges that are raised regarding framing (qualification) of the product. One specific task is to frame the product as a‘‘good’’, which means defining its functions and its potential for fulfilling customers’ needs. A ‘‘good’’ is a state and anobjectification of services valued by involved parties (Beckert, 2009). It is important to note, however, that the definition andvaluing of a good is a process where the good is singled out and differentiated in relation to other goods. However, there isalways the possibility of reframing. Chua and Mahama (2007), for example, illustrate how the parties in the relationschanged their views on value over time, thus creating controversy.

In relation to the aim of this study, the reasoning thus far helps us further specify its focus. While much of the mainstreamliterature takes the entities involved in HORs as stable a priories the framework presented here urges us to recognize theformative processes where actors and entities are formed and stabilized in relation to each other. The reasoning directs ourattention to the organizational arrangements that facilitate the process of framing. How relations, actors and products/services are formed and what causes changes become central questions to address. This means that the reasoning must bebased on an understanding that entities, often treated as stable and well-defined, are in fact unstable and contested(Mouritsen and Thrane, 2006). We must pay attention to formative problems and how a frame is obtained and maintained aswell as how the forming of actors and relations are affected by accounting and control initiatives.

2.2. Overflowing

While the concept of framing helps us identify the actions taken to stabilize the relations, the concept of overflow helps usidentify and discuss the incompleteness of the frame. Overflow directs our attention to the fact that framing is a process,which is never completed – there will always be overflows (Callon, 1998b). No matter how advanced the contract is, therewill always be loopholes. While the actors engaged in relationships strive to make the frame as complete as possible, theywill never fully succeed. Callon (1998b) argues that there are two types of overflows that we must consider in our analysis.The first source of overflow is to be regarded as caused by the actors’ inabilities to include everything relevant (or with thepotential to turn out to be relevant). The second source is the actions, techniques and structures used in the framing process,which are all potential sources of overflows.

First, in order to reduce complexity, which is a prerequisite for calculation, the relations have to be disentangled, or put inanother way; the outside world has to be ‘‘bracketed’’ (Callon, 1998b). But what has been put in brackets has the potential tobe actualized in the framing process if forced back into the negotiations. Callon provides examples of how externalities, suchas environmental problems, might be forced into the calculations and the framing process, if there are actors strong enoughto advocate it. However, unlike economists who often treat externalities as a market failure or a disequilibrium, Callonstresses that overflows are common and rather constitute the norm than the exception. Following this reasoning, it isimportant in the analysis of HORs not only to consider what actions facilitate interaction, but also how these actions build ondisentanglements, since this is a potential source of overflows.

Second, overflow might also be caused by the very measures taken to minimize the risk of overflows. The actors take avariety of actions and use different elements in the process of framing relations. Callon exemplifies with contracts,measurement instruments and definitions of roles. A contract, for example, ties parties together and makes it easier topredict the actions by other parties. However, the contract also presupposes the backing of a legal system and with that(maybe unwanted) transparency and publicity of certain dimensions of the relations if the parties take an unsettled issue to





G Model


court. When we conduct our analysis we must therefore also consider what relations are created in the attempts to frame theHOR.

In the next section, the role of accounting and control is explicated and it is argued that we can expect accounting andcontrol practices to both provide stability and to destabilize. We will develop the reasoning conceptually by tying accountingand control to the concepts of framing and overflow. As a matter of fact, based on the extant literature, we can expectaccounting to play a central role in framing HORs. When it comes to overflow the literature is scarcer, but we might expectaccounting and control to play a role in these processes as well.

2.3. Accounting and control as framing devices

One central theme within mainstream literature has been the focus on how accounting and control creates stabilizationthrough reducing the effects of goal incongruence, distrust, and misalignment (Das and Teng, 1998; Dekker, 2004;Vosselman and van der Meer-Kooistra, 2009). These studies illustrate the instrumental and tool-like nature of accountingwith stabilizing effects. Accounting and control are treated as tools that rational actors use in order to avoid risk and improvecoordination. Control-oriented studies with a non-reductionist approach have also shown that control helps stabilizerelations (Coad and Cullen, 2006; Dechow and Mouritsen, 2005; Kurunmaki and Miller, 2011), not only through riskreduction and coordination, but also through facilitating the development of the relations. Mouritsen and Thrane (2006)observed control activities that they categorized as orchestrating and that helped structure the relations and reposition theparties vis-a-vis each other and the network. Their study, therefore, indicates that control might be essential not only in thereduction of risk and improvement of flows, but also in the transformation process of HORs.

Accounting and control are, however, not only inter-mediating instruments in the sense that flows of information and thesignaling of needs for action are facilitated. Control is also a mediating instrument that in itself forms part of the translationprocesses (Boedker, 2010; Miller and O’Leary, 1987; Roberts and Jones, 2009). Accounting is not only about counting, but alsoabout the ordering of what is counted. Not least, control is based on ‘‘strong’’ numbers and figures that have the capacity totrigger consequences that are sometimes unexpected by the actors initiating them (Briers and Chua, 2001; Lowe, 2001). Thisviewpoint means that accounting is not fully, or even at all, in the hands of a principal. The latter is illustrated in Mouritsenet al.’s (2001) study of control initiatives in two HORs. The control initiatives (target cost accounting/open book accounting)were introduced in order to enhance the exercise of control at a distance and to create transparency. While these initiativeswere successful to some degree, the authors also observed other consequences. The control initiatives, when introduced andstabilized, also affected the interacting parties who changed their roles and their identities. The authors’ conclusions werethat accounting and control initiatives may be stabilized and have effects that are more far-reaching and uncontrolled thanexpected. This is also what Carlsson-Wall et al. (2011), in a departure from another theoretical position, observe in theirstudy of inter-organizational and intra-organizational control. Measures taken to control the relations betweenorganizations affected the intra-organizational control of the interacting organizations.

The referred studies indicate that accounting and control initiatives might facilitate a movement of a network of relationsinto a more dense state, in the sense that the interacting entities rely on and become more dependent on each other(Wagenaar, 2007). However, Mouritsen et al. (2001) stress the importance of recognizing that mediating techniques, onceintroduced as a tool by one actor, may have far-reaching consequences that shift power and strengthen the network. Controlmay be no one’s ally (Lowe, 2001). We can expect that several control initiatives will be used simultaneously and thatdifferent control initiatives have different consequences as they change status over time (Chua and Mahama, 2007; Dekker,2003). This draws our attention to the temporal nature of stabilizing effects.

2.4. Accounting and control as a potential source of overflow

In order to understand accounting and control in the framing process of HOR, the analysis must consider the attempts tominimize risk and to enhance flows as well as to recognize the dimension of control as a force of its own – a force with thepotential to strengthen and integrate HOR in a process where no obvious party except the HOR is distinguishable as initiator.Still, the reasoning only highlights the stabilizing effect of accounting and control. By explicitly discussing accounting,control and overflows, a theoretical foundation for analyzing potential destabilizing effects is provided. We identified twopotential sources of overflows in the discussion on overflows in the previous section and here we tie the reasoning toobservations made in the accounting and control literature.

Firstly, overflows might be created through the disentanglements that accounting and control contribute to. Because ofthe many possible connections, reductions must be made (Callon, 1998b). It is impossible to include everything in the frame,and accounting and control initiatives facilitate the exclusion of complexities. The boundary creating and maintaining therole of accounting has been stressed (Llewelyn, 1994; McNulty and Ferlie, 2004) and it has been shown how accounting canisolate and enhance certain connections (Lowe, 2001). Accounting and control might therefore enhance the HOR and make itdenser at the cost of the ability to connect to other actors and relations. The integrative effect of accounting also becomes apotential source of overflows. The introduction of various control initiatives in order to handle uncertainty creates a ‘‘hypercomplex’’ situation (Luhmann, 1995). Systemizing and organizing flows of information rests on processes of reductions,which in turn create ‘‘blind spots’’ (Dechow and Mouritsen, 2005). However, what has been externalized has the potential tobecome problematized and actors might push for a reframing (Callon, 1998b).





G Model


Secondly, accounting and control activities might, in themselves, open up the frame. The attempts to disconnect may leadto enhanced complexity through new connections and openings to other ‘‘worlds’’ (Chua and Mahama, 2007). Strathern(1999) illustrates the point when discussing patents. While a patent is filed with the intention to safeguard property rightsand make it possible to calculate on possible gains and losses; the patent process also contributes to publicity and visibility.Chua and Mahama (2007) also show how accounting and control might for example connect actors to specific issues that hadearlier been separated, resulting in controversies. Attempts to isolate may therefore lead to reconnections to unexpected‘‘worlds’’; what has been disentangled is reconnected and accounting and control might facilitate this.

Although the reasoning is tentative, all in all we can expect accounting to cause potential overflows, both through theintegrative functioning making relations denser and through unintended reconnections; through establishing connectionsand through cutting them off. This conceptualization provides a potential explanation for why control sometimes onlyfacilitates stability for a limited period of time (Chua and Mahama, 2007), and then ceases to have that effect. However, wemight also expect stability and controversies to be produced simultaneously.

3. Implications of the framework

The theoretical point of departure in the core concepts of framing and overflow helps direct attention to the problems ofHORs, while not beforehand defining and deciding on causes. One important observation is that actors, relations, controlmodels, and products/services must be understood as constructed. Consequently, if the aim is to better understand HORs andthe core problems of these organizational arrangements, our attention must be directed toward framing attempts andoverflows. While much of the focus in previous research has been on initiatives that create stabilization and on the effects ofsuch stabilization attempts, less attention has been directed toward disentanglement and its results. Accounting and controlactivities might not only be an essential instrument in the framing process but also a source of overflows.

The framework has guided the structure of the analysis that follows the empirical descriptions of the case study. Theanalysis centers on three themes:

(1) A

Pl1

discussion of what was problematized in the relations. The theoretical framework directs the attention not only toproblematizations regarding flows, risk and opportunism but also to the formation of actors, relations and products.

(2) T
he description of the framing process with a focus on what actions and measures were taken in order to stabilize theHOR. A central issue is the way in which accounting and control was used in order to frame the relations.
(3) A
discussion on the sources of overflows and more specifically on how these might be connected to accounting andcontrol initiatives. The analysis centers on the extent to which accounting and control created ‘‘blind spots’’ and on theextent to which accounting and control reconnected the framed relations to other worlds.
4. Empirical setting and approach

One difference between most other studies of HORs and this study is the empirical setting. This study is set among variouspublic sector organizational units that cooperate around a shared service center, which in turn provides IT support andadministrative services. Both the shared service center, as an organizational solution, and the public sector have specificcharacteristics that make them suitable for the theoretical approach taken in this study. Our theoretical approach stresses theimportance of identifying processes of formation rather than of states (Mouritsen et al., 2010). The shared service center is oftendescribed as a stage in the movement toward a ‘‘true’’ inter-organizational arrangement (Janssen and Joha, 2008). Thus, theshared service center is a setting in which the processes of framing and disentangling services and units, and establishing andmaintaining relations, may be observed. Also, the public sector, where goal incongruence is common and where there are fewertangible products, constitutes a good setting for identifying not only relations and exchange arrangements (AkerstromAndersen, 2000; Fountain, 2001), but also the processes of defining the services produced (Orlikowski and Iacono, 2001).

This is a longitudinal case study in which relations were observed as they happened. This approach led to an analysis ofhow these relations developed. This is in line with Mouritsen et al.’s (2010) argument that stresses the importance offocusing on networks of organizational relations ‘‘in action’’. In addition, several scholars (Caglio and Ditillo, 2008; Dekkerand Anderson, 2010) have argued that there is a lack of and, consequently, a need for longitudinal studies of control in HORsince such studies provide an empirical foundation for understanding the dynamics of HOR.

4.1. Introduction to the empirical setting

The empirical study shows how the relations evolved around a shared service center in a large municipality in Sweden(about 500 000 inhabitants) during a three-year period. The shared service center (the Provider) mainly delivered services tothe 21 municipal districts and to some technical departments as well as to external customers (although this ceased). Eachgeographical district provided the same kind of services within their geographical area including larger services such asschools, elderly care and social support.

Each district had its own political board. According to the annual statement for 2008, the smallest district’s total spendingwas 460 m (SEK) and the largest district had a total spending of 1282 m (SEK). Each district had their own head of district andstaff (ranging from all functions like finance officer, HR officer and heads of the different services). The headquarters of the

ease cite this article in press as: Kastberg G. Framing shared services. Crit Perspect Account (2014), http://dx.doi.org/0.1016/j.cpa.2014.01.002




G Model


districts were located in the districts. The Municipal secretariat was not a large customer but, as will be described in moredetail further on, took on a coordinative role. The interviewees described the organizational model with districts asdecentralized in that the districts had authority over most questions regarding the operations of the services. This alsoincluded decisions regarding IT and administration.

The Provider had its own board. The majority of the members were representatives of the users. However, two of theboard members were external and had been appointed because of their special knowledge regarding IT and shared services.It was the municipal council that appointed the members of the board.

4.2. Empirical foundation

The study is based on empirical data from different sources. The main empirical sources are the 65 interviews conductedin three different time periods. The first study was performed in late 2006 (25 interviews), the second in early 2008 (15interviews), and the last in the Spring of 2009 (25 interviews). In each part study the respondents were asked not only toreflect on the current state but also to describe what had happened during the last period of time.

Respondents representing different parties were selected, covering representatives for the Purchaser, the Provider, thedistricts/users and the Provider board. The selection of respondents was not fixed at the beginning of the empirical studies inorder to keep openness to new potential respondents. However, one aim was to return to the same respondent in the threepart studies. One reason for this was pragmatic; most of the respondents were key persons in different positions. Returningto the same respondents also allowed for a deepening of the questions since the last interview formed the foundation for thecoming interview.

In addition to the interviews, meetings were observed and documents studied. In total, four meetings of the Providerboard were observed, the purpose being to gain insight into what kinds of questions were discussed and in what way. Theresearcher took on the role of a silent and passive observer. The board meetings had a formal character, which probablyreduced the disturbance caused by an external observer. Documents were also studied, like financial statements covering thefocal years as well as statistics, which helped nuance and validate claims made by interviewees. Several investigations hadbeen carried out by different involved actors before the study was initiated. These documents were especially valuable forreconstructing the development before the first interviews were conducted.

The interview guide used was adjusted to suit the position of the respondents and also in order to take advantage of theresults of empirical descriptions that had been generated. The questions were open-ended and there was flexibility asregards the order in which the questions was asked. Typically, the interviews started with a broad question about how therespondent perceived the current situation. What issues and problems were at the top of the agenda at the moment and whatactions were discussed? Another typical question in the first phase of the interview was to ask the respondent to reflect onhow things had evolved during the last years. Then more detailed questions followed regarding the relationship to differentparties and how different control measures had developed.

5. Shared service center relationships

This section presents an account of the development of the relations around the shared service center. The structure of theempirical descriptions mirrors the longitudinal character of the study in which problems and control solutions together withthe accounting and control initiatives are described as they evolved over the years. Each section concludes with a summaryof the core problems and the measures taken.

5.1. Before 2006: revised relations and a new governance model

The first interviews were conducted in 2006. However, documents and the respondents’ accounts provide a picture of theproblems the actors struggled with prior to 2006. When the first interviews were conducted, a new model for interaction hadrecently been introduced and the actors referred to it as the Governance model. However, in order to understand the modeland why it was introduced, it is necessary to look at the years prior to its introduction. As Chua and Mahama (2007) conclude,it is important to take the history of the relations into consideration.

Since the 1980s, the Provider had provided IT solutions to the 23 municipal districts, technical departments, and externalcostumers. The Provider had its own Board of politicians. In terms of accountability, the Provider functioned as a profit centerand was obliged to show a profit. Each municipal district negotiated directly with the Provider about services, price, andquality. According to the respondents, the Provider was very autonomous, business-like and user focused. Custom-tailoredservices (e.g. IT systems and support) were offered to meet the needs of the users. This resulted in a number of different ITsystems in use. But, as several respondents emphasized, this was an expensive situation.

Plea10.1

And it was like a place where you could wish anything. The persons sitting there were the persons working with theseissues in their everyday work. They could order anything that was related to their user situation. It becametremendously expensive. It was this that we wanted to change. (User representative 1.)

There was a sense among the users that the costs were uncontrolled and increasing. Among the districts there was anexpressed dissatisfaction with the situation. Actually, users had no idea about the total costs because of the lack of overall

se cite this article in press as: Kastberg G. Framing shared services. Crit Perspect Account (2014), http://dx.doi.org/016/j.cpa.2014.01.002




G Model


coordination among them. There was great distrust between the users and the Provider. The users thought they could nottrust the Provider to act in their best interests while the Provider thought this distrust was misdirected and unfair. Inresponse to this situation, the municipal districts began to organize as one purchaser with the ambition of acquiring controlover costs.

Plea10.1

From the perspective of the districts I believe that they felt that they were at a dead end. They paid too much. It’s allabout merging into one Purchaser with one body of competence, or else they will not make it. They will not make it inrelation to the Provider, and they will certainly not succeed in making it on the market. (Municipal secretariatrepresentative 1.)

A first step was to form a group with appointed members from the districts. The users, now cooperating in order to gaincapacity to act, started to investigate alternative solutions, one being outsourcing and another being the reframing of theexisting relations with the Provider. One important insight in this process that was stressed by several intervieweesrepresenting districts and the Municipal secretariat was that they lacked enough knowledge to choose the market option:

We have said that in this large city, we don’t know what’s out there in the bushes – the IT systems and supportavailable. We have also said that we can’t outsource before we know what we are outsourcing. We have no idea ifoutsourcing would be cheaper or not. We must work together; otherwise, the market will devour us. (Municipalsecretariat representative 2.)

The districts also conducted a benchmarking analysis of the costs of the IT support and administrative services; theyconcluded that they were paying too much. The benchmarking was carried out by external consultants. The Provider,however, felt that the results of the benchmarking were incorrect and conducted its own benchmarking showing that thecosts of their services were not too high. Consultants hired by the districts and the Municipal secretariat performed several,quite extensive investigations with the goal of identifying problems and recommending possible solutions. It was suggestedthat the relations should be reorganized in a more efficient way and the idea of the new Governance model was introduced.The idea of the Governance model was also supported by the Municipal secretariat.

The Municipal secretariat had other concerns and interests than the districts. While the users were mainly concernedabout costs, arguing for change using the slogan ‘‘more IT for the same buck’’, the Municipal secretariat saw an opportunity tostrengthen its oversight of IT operations and to raise security issues. This should be understood as stemming from a situationwith a highly decentralized organization with little coordination between the districts and where the Municipal secretariathad a reduced role.

And we could observe that we could not follow up since they (the districts) ordered and developed their own systems.That made it hard for us to grasp the totality of the municipality. We must of course know how elderly care is developingregardless of the fact that the elderly care is organized in 21 districts. (Municipal secretariat representative 1.)

The core idea of the Governance model was to establish well-defined roles where the three most significant parties werethe users (municipal districts and technical departments), the Provider, and a new Purchasing unit. In addition, new boardmembers of the Provider board would be appointed. User representatives, acting in the best interests of the users, wouldreplace the politicians on the Board, who had acted as ‘‘owners’’ of the Provider operations. The Purchasing unit was anextension of the group that the municipal districts had formed earlier, but the unit was located at the Municipal secretariatand was headed by the IT-director.

Besides these changes, there were several other components in the Governance model. Several measures were taken tomake certain that the Provider did not fall back into the role of a seller who develops and provides specific systems and tomake certain that the users did not order tailor-made systems. One measure was to introduce the concept of the ‘‘one sharedsystem’’. This meant that all users would use the same systems, without adjustments. There should for example only be oneHR-system. The aim was to realize synergies. Users, in negotiations in user forums, would decide on the specific featuresdesired in the shared systems; these features would then be formalized in orders from the Purchaser (the new purchasingunit):

When we have defined a shared system, we allocate all wishes and demands via the person with us who is responsiblefor that system. After that we make a plan and place an order. This is instead of all users making their own orders withthe Provider. (Purchaser representative 1.)

It was clearly stated that no direct contact on orders should take place between the users and the Provider. Allcommunication regarding the dimensions of the systems should go through the Purchaser. In order to promote the rightincentives, the Provider was a cost center in the Governance model; the Purchaser would stipulate all costs in the orders itprepared. A cost allocation model was introduced where the costs for the services were split among the users on the basis ofusage. To keep costs reasonable, an annual benchmarking analysis was required.

The introduction of the ‘‘one shared system’’ was also important because the situation with the many tailor-madesystems had unwanted side effects. One such side effect was that the users became extremely dependent on the Providerbecause only the Provider had the competence to maintain and develop the systems. One clear aim was therefore to movefrom a situation with custom-designed systems.





G Model


5.2. Summary of observations

When reviewing the events before the first part of this study several problems triggering actions can be identified. Manyof the problems regarded a weak framing of actors and products in the sense that interaction was hindered or problematized.The product and services provided is one central example of this where the municipal districts as buyers and users found itproblematic that there was such a wide range of services delivered without any coordination. The problem of lack ofcoordination was also perceived by the Municipal secretariat, but more regarding the lack of overall view and control. Otherdimensions of the relations that were problematized were the role and identity of the actors, like the profit seeking Providerand the weak role of the users as purchasers. The descriptions indicate that the key actors reacting to the circumstances werethe municipal districts and the Municipal secretariat. The key measure initiated was the introduction of the new Governancemodel, which was much about redefining roles, defining the products and services and introducing various control measureslike a new cost allocation system and benchmarking.

6. 2006–2007: relations become more dense

According to the empirical descriptions, the implementation of the Governance model took several years and when theinterviews were conducted in 2007 there were several issues that the actors struggled with. One recurrent problem was thedefinition of roles and responsibilities, which became evident in different ways.

Plea10.1

The Governance model is quite theoretical. We have discussed it several times in a group, and what we have done asmembers responsible for the board is to interpret the assignment and see what consequences it has. (Member of theProvider Board 1.)

The Governance model stipulated new roles for the actors. The new role of the Purchaser had to be introduced since it did notexist before. However, it took time to develop that role and to recruit staff with the right competence. For the Provider, the initialphase of the new Governance model meant a lot of work with redesigning the internal organization so as to meet the newcriteria. Since the Provider was required to cease developing new systems, it no longer needed personnel with the competenceand skills for that work. The Provider also had to accept that its role was that of a competitive provider who provided only whathad been ordered. In order to safeguard competitive prices, much emphasis was placed on increasing efficiency. Also, the newlyappointed Provider Board members struggled with their conflicting roles–they were the representatives of the Provideroperations, but they were also the users’ representatives with a close connection to the users’ core operations. For example,sometimes the work associated with implementing the ‘‘one shared system’’ meant that the Board members, against their will,had to promote certain shared system solutions without the adjustments needed by individual users.

Beside the definition of actors and roles, also products had to be defined. One problem during the initial period wasdefining what the ‘‘one shared system’’ actually meant. The users had several IT systems, some more specific to individualuser operations than others. This raised the question of which systems would be shared and which would not.

The relations between the different roles were also problematized, especially the relations between the Purchaser and theProvider.

According to my opinion, what has been an issue throughout the year, has been a lot of work, and a lot of discussionsregarding our cooperation with the Purchaser. I mean, in order to generate some kind of structure and control of therelationship. Actually, this still does not function in a good way. (Provider representative 1.)

According to the Governance model, the relations should rest on a high degree of formalization where the Purchaserplaced formal orders with specifications about both services and costs. This idea about the relations was described in thedocuments, and interviewees claimed that this was done so as to safeguard that only what was really needed was ordered. Infact, in general, the communications were less formal. One problem was that, because of the novelty of the situation, newquestions were constantly raised. In addition, the Purchaser had a low capacity caused by understaffing and lack ofcompetence in placing orders. The Purchaser therefore frequently had to turn to the Provider and initiate a dialog onproblems since the Provider had the competence to deal with such problems. However, these dialog-based relations werequestioned, not least by users’ representatives on the Provider’s Board, who argued that the model did not intend thissituation. The problems of role confusion and the lack of well-functioning communications are illustrated in the followingtwo researchers’ notes from a Board meeting:

One Board member asked who placed the orders. The technical department manager answered that the Purchaser did.Both the Board members and the IT director said this seemed strange because this was outside the responsibilityspecified by the Governance model. The IT director said it was the Provider Board that made such decisions. A Boardmember said: ‘‘We must make the decisions on what can be ordered’’.

The problems of responsibility and the arrangement of the relations led to discussions in different forums and to theinitiation of ‘‘higher meetings’’ with representatives from the Provider Board, the Purchaser, and the Municipal secretariat.The aim of these meetings was to resolve these problems.

Another problem for the Purchaser was the establishment of well-functioning channels to the users. According to theGovernance model, the Purchaser placed orders for the users. However, the model assumed well-defined definitions of user





G Model


needs, which was not the case. Although the work of establishing user forums was initiated, during the first two years theresults were unsatisfactory. It was difficult to spread the idea of the forums among the users. In order for the model tosucceed, this was regarded as essential. A consequence of the lack of information was that orders were not anchored amongthe users, leading to doubts about the arrangements since the delivered services did not meet their expectations.

A great deal of effort was also required in the first year of the Governance model in order to establish the new costallocation model. Procedures had to be set for the cost allocations, and the ordering cycle had to be synchronized with theoverall budget process in the municipality. Another control initiative was the annual benchmarking performed by anexternal consultant. All parties regarded the results of the benchmarking as important information that indicated whetherthe Provider’s prices were consistent with market prices. If there were problems, the market alternative was an option.


During the period we can observe how the actors struggled with establishing an interaction in accordance with theGovernance model. Although the model was simple and straightforward in theory, the actors problematized their acting inrelation to the model and, in addition, some roles were introduced as new. There were also question marks regarding thedefinition of the products and services. Besides, the struggle with establishing the entities in the relations was problematized.For example, it was hard for the Purchaser unit and the Provider to keep up a distanced and formalized relation.

Altogether we can see that the actors had to put a lot of energy into forming the relations in a way that supportedinteraction. Internal organizational activities were initiated in order to meet the demands placed by the Governance model.The Provider for example had to change its competence and ‘‘reinvent’’ itself as a mere Provider of what had been defined andordered. The problem with definitions of products and services led to activities as well, such as intensified work withestablishing the user forums. Besides that, benchmarking activities and formal control such as the cost allocation model wereintroduced, although the latter took longer than expected by persons involved.

7. After 2008: dense relations

After the first two years of struggle, the relations started to work more in accordance with the new Governance model.The Provider had redefined its role and acted as a provider of services only ordered by the Purchaser. However, the Providerwas criticized for being too passive and for too often passing problems on to the Purchaser. Interviewees representing theProvider on the other hand argued that this was in line with the Governance model. The Provider Board members no longerproblematized their contradiction as Board members and user representatives, as they had earlier. They became more‘‘professional’’. The Purchaser was better staffed, but still lacked some competences needed for its task:

Plea10.1

They (Purchaser unit) are still not fully staffed. Their assignment is too large in relation to how many they are. They aresupposed to work against the operations in the districts and to have contact with the user forums and with us asproviders. They are still not up and running. One good example is how the agreements look where our assignments areformulated. It should be specified in service level agreements, but in fact it is very non-coherent. (Providerrepresentative 1.)

Some older problems remained, however, and new problems had arisen. One new problem resulted from theimplementation of the cost allocation model. Before its introduction, there was insufficient knowledge on the actual costsituation for IT support and administrative services. With the new model for costs, the greater transparency led to newquestions since it was possible to see not only the total purchasing costs, but also the fact that such costs were increasing. Forthe users, constantly under pressure to perform under tight budgets, the increasing costs were seen as something that ‘‘stole’’resources from their core operations such as schools and elderly care. However, the Purchaser’s representatives argued that,in fact, total purchasing costs were not in themselves a problem since the actual issue was the costs in relation to each order.If every order were motivated from the perspective of the user who had ordered it, the total purchasing costs were not ofinterest. If the introduction of a system application actually saved money, then an increase in total purchasing costs wasbeneficial. However, as it was, the quarrel about total purchasing costs led to skepticism about the value of the HOR and thefunctionality of the Governance model.

Moreover, once the idea of the ‘‘one shared system’’ was implemented, the question of which systems should be sharedwas resolved. However, there were different reactions from the users. For example, users in the districts began to questiontheir own way of working. With the introduction of a standardized system, it was noticeable that the operational routinesand procedures differed among the users across the districts. As a consequence of this observation, work began onstandardizing processes so that the various users would work in the same way in ‘‘shared processes’’.

What kind of processes do we have that are the same? And then we identified recruiting, skills development, wagesetting and rehab, that kind of large processes came up. After that we have had process groups with HR managers (ofthe districts) who have been working with all the issues and agreed that these processes we own together andtherefore we should work in standardized processes. (User representative 2.)

Reactions in the opposite direction could also be observed. The introduction of the ‘‘one shared system’’ also caused usersto question whether it was reasonable to assume that all systems operated under similar conditions with similar system





G Model


requirements. Standardization seemed to have gone too far. One example that several respondents referred to was thespecific conditions at the schools, where there was a cost for each computer. This cost was calculated based on average usageby a municipal employee. However, since many of the users in the schools were students, this cost model was inappropriate.The students amounted to a large user group in relation to the districts’ personnel and the students had other needs. For thepersons responsible for the school budget this became costly and they argued that there were other, cheaper, solutions forschools than the ‘‘one shared system’’ for PCs. Their argumentation seemed successful and when the studies were carried out,several respondents representing different perspectives argued that there had to be a change. According to the respondents,a new solution was planned for the cost arrangements for computer usage at the schools in order to suit the specificconditions.

Plea10.1

The students are supposed to be able to use the devices and opportunities they have privately. You know, to be able touse it in school and combine it instead of having two systems. And of course, it is also a matter of costs. We are notprepared to tie up so much on IT if it does not provide us with possible teaching gains. So, we are opting for anothersolution where we look at one-to-one solutions, a little bit of Apple, some other alternatives, and now we also open upfor wireless solutions. (User representative 3.)

Another problem was that the user forums still did not work as intended. Although most user forums had beenestablished, they still did not channel needs and anchor the orders. This had several consequences, one of which was that theorders placed were too costly. The respondents said that the high costs were a result of the fact that users appointed people tothe forums who were concerned only with requirements, not costs. The inability of the users and the Purchaser to channelthe orders also led to confusion among the users about what to expect from the services provided, causing a gap betweenexpectations and results. Thus, the situation was troublesome for the Provider who feared being blamed (again) for providingservices not ordered and for doing so at too high a cost.

The maybe most important question that the Purchaser and the Provider have in common is to identify theresponsibilities. . . we have worked a lot with this. We have reached far but still there are issues to deal with when itcomes to describing the organization, what the services are, what can be expected regarding the supply and how theycan affect the content. . . who is communicating what, this is a real challenge in a large organization like this. (Purchaser

representative 2.)

As a reaction, the Provider pushed for more formalization of orders and for the introduction of Service Level Agreements(SLA) that would specify exactly what had been ordered and consequently what the users could expect.


In relation to the previous part studies we can see that some problems remained, although some of the thingsproblematized were new and yet still others were problematized in a new way. One of the remaining problems was thatthere still was uncertainty about the definitions of the products and services. One identified contributing factor was thefailure in establishing well-functioning user forums. It was hard to make the forums representative of different stakeholders’interests and the needs formulated by the forums were described as unanchored in the districts. Several of the new thingsproblematized were reactions to the new way of interacting in accordance with the Governance model. Examples of suchproblematizations regarded reactions to the standardizations of the services and to the fact that the Provider was playing itsrole too well; it had become too passive. The now functioning cost allocation model also caused problems since it generatedtransparency regarding the costs that, in turn, led to problematizations.

We can also observe how several measures were taken in order to improve the relations. One such measure was to startinvestigating how to depart from the idea of ‘‘one shared system’’ when necessary, but also how to develop the operationalroutine across the municipal districts in order to better make them ‘‘fit’’ into the systems. The remaining problem with theuser forums was met with actions addressing the question as to how to make them representative, but also how to formalizethe orders to a larger extent in order to create well-defined products and services.

8. Analysis

In the literature, HORs have been described as arrangements that are pulled together where different mediatingtechniques increase the tightening of the relationships (Kurunmaki and Miller, 2011; Mouritsen and Thrane, 2006). At thetime when the first empirical study was carried out, the studied HOR were loosely coupled and fragile; actors argued forbreaking up the relations. However, the empirical case descriptions indicate that the HOR became denser over time. Thesemovements will be discussed in more detail in the analysis, starting with a discussion on the issues that were problematizedduring the study. After that we will turn to the framing attempts and the observed overflows.

8.1. Formative problems – lack of stable frame

We begin our analysis by addressing what the main issues were and what the actors trying to form the HOR struggledwith. Following the theoretical framework, we pay special attention to problems related to the framing of entities and





G Model


relations. What we observe is how the interaction was problematized and how overflows were produced as a consequence ofthe lack of a stabilized frame.

The empirical descriptions illustrate a range of risk reduction and coordination problems, which the actors, when tryingto interact with each other, stumbled upon. The problems of cooperation and coordination that are often addressed in themainstream literature (Caglio and Ditillo, 2008) were also evident in our case study. The actors questioned other actors’motives, which triggered attempts to minimize vulnerability and the introduction of for example a transparent costallocation model, benchmarking and a cost accountability model governing the Provider. But there were also problems of amore coordinative nature connected to the problem of securing flows (Tomkins, 2001), such as defining how soon ITpersonnel should answer calls from users. The described problems are, as indicated, in line with the challenges most oftendiscussed in the mainstream literature. Still, as the empirical descriptions show, a lot of difficulties were of anothercharacter. The process of framing HOR is not only about framing relations between stable entities but also much aboutframing and stabilizing these entities.

Throughout the study we could observe several problems because the actors, relations, and products/services were, infact, not defined or established, which is in line with the observations made by Mouritsen and Thrane (2006). There was alack of a priories that had to be decided on. In line with the discussion in the theoretical section, we label them formativeproblems. New identities in relation to others had to be established and services had to be defined. The Governance modelfollowed several years of discussions and investigations, and contained specifications on roles, relations, and the ‘‘product’’.When scrutinizing what happened before the Governance model was introduced, the descriptions of the development wereto a great extent a problematization of the existing roles and relations. One of the core problems described was, for example,the role of the Provider, who had been too profit-seeking and customer oriented. Other problems were the uncoordinatedrole of the users and the much-reduced role of the Municipal secretariat. The empirical descriptions also show how theformation among the users into a more united and well-organized actor shifted the allocation of power and led to theinitiation of what eventually became the Governance model.

Although the existing definitions of roles, products and relations had been problematized, the Governance model solutionand its specifications did not solve the struggle. Throughout the study we could observe how the formative problemsrecurred. One example of this is how the Provider’s role, as a provider only of what had been ordered, by the end of the studywas problematized and described as too passive. Another problem regarding a role was that of the Provider Board where themembers, at least initially, struggled with the conflict between on the one hand being a representative of one specific userorganization and on the other hand representing all users. A third example is the Purchaser, who throughout the studystruggled with gaining capacity, both regarding staffing and competence, in order to be able to function as prescribed in theGovernance model. Still, it was not only the actors’ roles that were problematized. Another example of a central formativeproblem was the qualification of the services. We noted the problem of the definition of services not only at the beginning ofthe study but also throughout our study. The problem of defining (that is, deciding on) the services reflects Callon et al.’s(2002) descriptions of how problematic it is, in practice, to qualify a service as a good.

Altogether, when scrutinizing what was problematized we can see how, besides the problematization of risk andcoordination, there were several recurring problems of formative character. Core problems regarded definitions of theservices, identities (for example profit seeking or not?) and responsibilities (who is representing who?). The lack ofagreement upon roles and definitions both functioned as a source of overflows and made the relations more fragile andcontested. As we will see in the next section, where we turn to the framing activities and the role of accounting and control,the formative problems triggered actions.

8.2. Framing initiatives and a more stable HOR

Despite the described problems, and as indicated in the introduction to the analysis, the situation developed into one thatcan be described as denser and the actors were pulled together. In the process of framing the HOR, accounting and controlinitiatives were central mediating devices facilitating coordination and risk reduction as well as an orchestration of therelations.

The introduced Governance model was a central mediating technique. The Governance model introduced not only newroles for existing actors but also new organizational units. Furthermore, much effort was spent on reallocatingresponsibilities. One major shift, for example, regarded the responsibility and the obligation to decide on what servicesshould be ordered. The former solution, with direct negotiations between all users and the Provider, was supposed to bereplaced by cooperation between user forums and the Purchaser. For the Provider, this meant reduced responsibilityregarding for example the development of services and decisions regarding service levels. The Provider Board also had newresponsibilities and did not only act as a representative of the Provider operations but also as a user representative. Theseframing activities, despite the struggle with actually living up to the roles described in the previous section, helped define theHOR. Deviations from the Governance model led to actions such as ‘‘high level’’ meetings about boundaries and definitions.What we observe is therefore how mediating techniques were important in the orchestration of the relationships, which is inline with Mouritsen and Thrane’s observations (2006).

We can also observe how more traditional accounting and control models functioned as mediating techniques thatfacilitated the integration of parties through risk reduction and coordination. One example is the cost allocation model withthe intention of facilitating the transformation of the Provider from being a profit center to a cost center. As a profit center,





G Model


the Provider’s role enhanced its separation from the users. There were doubts among the users as to the motives of theProvider; whether it worked in the user’s best interests or only to seek profit. One clear purpose of the cost allocation modeland of the Provider being a cost center (besides its obvious function as a cost allocator for the users) was to reduce theProvider’s incentive to take a ‘‘selling’’ position by offering new (costly) solutions. The change in financial responsibilitytherefore both contributed to the reduction of risk for the users, and to the redefinition of the identity of the Provider.Another control attempt that had a potentially integrating effect was the introduction of SLA:s. These agreementsconstituted an essential initiative in the process of defining the products and in the attempt to reduce the quarrels aboutorders and deliveries. Alongside the framing activities of a more organizational character, we therefore also observed severalaccounting and control initiatives and how they enhanced interaction through risk reduction and coordination as well asthough facilitating redefinitions of roles.

The process of the creation of a dense HOR is also a process of disentanglement. Following the theoretical framework, it isimportant not only to pay attention to the framing activities, but also to the disentanglements that these actions cause. Theintroduction of the ‘‘one shared system’’, for example, rested upon active framing activities which, when scrutinized, alsorevealed disentanglements. It was not just about reducing the number of systems. The variety of systems to some extent reflectedthe complexity of the operations in the municipality. Different sectors had different demands. In addition, to some extent, thesame sectors in different geographical areas operated under different conditions. With the ‘‘one shared system’’, all user needscould not be considered. The introduction of the ‘‘one shared system’’ therefore also meant reducing the embeddedness of thesystems and making them sufficiently universal to suit all users. The ‘‘one shared systems’’ were disentangled from thespecificities of the operations among the users. There were other examples as well, such as the establishment of a new model fordeciding on what services should be provided, which constituted an attempt to reduce the opportunities for the users to ordercustom made solutions. The organizational solution, with one Purchasing unit through which all orders should pass, also had theeffect of disembedding the relations from the local needs and the specificities of the different users in the HOR.

The description of the process moving toward a denser, more disentangled HOR is also a description of a shift in power.Using the terminology of Mouritsen et al. (2010), the network started to ‘‘bend the world around it’’. One example was theinitiation of adjustments in the operations of the users. The notion of the ‘‘one shared system’’ triggered discussions aboutthe multifaceted operations performed at different locations. If the systems were to be standardized, then the processes of atthe operations had to be standardized in order to ‘‘fit’’ the standardized systems: ‘‘one shared system’’ was accompanied by‘‘one shared work process’’.

In the reallocation of responsibilities, a problematic issue was how to deal with the definition of services: what should bedelivered, at what price, and with what quality? When the relations were established and the various roles defined, theproblem of defining services and costs was externalized in the sense that it was defined as a user issue. If services were tooexpensive or did not meet expectations, that was because the users were unable to channel the right specifications throughtheir user forums. The HOR had gained the ability to decide which problems were ‘‘inside’’ the framed relations and,consequently, which problems were ‘‘outside’’. With this position of power, the HOR no longer functioned as an ally of asingle actor. The identities of all parties were called in to question with reference to the Governance model, including theusers who had initiated the model. This is in line with Mouritsen et al. (2001) who also observed how the HOR became morepowerful, which resulted in changes in identity of the interacting parties.

The process of establishing denser HOR was a process of framing and disentangling. The result was a more powerfulnetwork which, at least to some extent, ‘‘bent’’ the world round itself. Like other studies (Coad and Cullen, 2006; Kurunmakiand Miller, 2011), this study also shows how different mediating techniques help form the network and make it denser. Theanalysis in this study indicates that the stability and the power of the system are partly results of the mediating techniquesused. However, our analysis also shows that the framing process rests on the pulling together of entities as well as on thedisentangling and dissolving of relations. In order to become more powerful and enable action, entanglements have to end.

8.3. Accounting, control and overflows

In the theoretical section we stressed the importance of scrutinizing the role of accounting and control not only aspotential facilitators in the framing process, but also with the potential to cause overflows. We have already described howthe denser relationships rested on a disentanglement, enhanced by the mediating techniques in use. This observation makesit central to address the extent to which the cut off links led to controversies, but also whether accounting and control led toreconnections.

One example of how accounting and control initiatives in fact reconnected the disentangled HOR to wider complexitieswas the cost allocation model. As described above, the costs were decided when the users defined their needs and thePurchaser translated them into formal orders. Among the users, the process was decentralized so that each operation (forexample, personnel, administration, schools, etc.,) defined its own needs. The costs were therefore decided in relation to thebenefits appreciated in each specific operation. Earlier on there did not exist a clear overall view of the aggregated totalpurchasing costs, but with the new transparent model such knowledge was gained. This caused controversies as described inthe empirical section. Some argued that it was of no interest to pay attention to total IT costs as long as users’ needs were met(maybe an increase in IT related costs led to lower costs in for example elderly care). Others, not least the people handling theusers’ budgets, called the total purchasing costs into question, leading to questions being posed concerning the nature of therelations and whether all parties were working in the same direction.





G Model


Therefore, what we observe is how accounting and control attempts also caused overflows. Cost controls both facilitatedcost allocation procedures and defined the Provider’s role, as described above. This increased transparency and reduced theperceived risk connected to interaction. Still, what we also observe is how cost controls created tension in the framingprocess by actualizing the complexities facing the people responsible for the users’ core operations. With the new costallocation model, the costs for IT and administration became visualized and, as an effect, related to the cost of the coreoperations. When the cost allocation model reintroduced these complexities, representatives of the operations forced theproblems back into the framing process. This illustrates how an initiative (allocation of costs) aimed at strengthening theframe can create formative problems regarding roles, relations, and allocation of responsibilities.

Besides the observation that accounting and control were sources of overflow caused by the connection of what hadpreviously not been connected, we could also observe how the disentanglements enhanced by accounting and control alsotriggered overflows. One source of constant overflow, at least during the period that the HOR was studied, was thequalification of the product. The concept of the ‘‘one shared system’’ filled an important function in the process of qualifyingthe product and also, as described above, triggered actions in the operations, which had to adjust to the new situation.However, not all operations adjusted. For example, for the schools the costs of using the computer services were too highbecause of their different needs. When introducing the concept of ‘‘one shared system’’, the multifaceted nature of theoperations became a blind spot, which in turn caused reactions. A discussion resulted that, during the last phase of theempirical study, suggested that a change in the ‘‘one shared system’’ for computer services was imminent. The multifacetednature of operations, which could not be excluded from the framing process, explains why it was so hard to avoid overflows.A system with several users rests on the logic of simplifications/generalizations whereas the needs of the operations arespecific. The problem of qualifying the product is also mirrored in the discussions and conflicts about the services, whichcontinued throughout our study. The Provider’s push for specifications in the SLA, in an attempt to further specify theproduct, indicates how the problem escapes the framing attempts.

The two examples of overflows illustrate how overflows were caused in different ways by the techniques used to framethe relations. Accounting and control contributed to problematizations and controversies both as a consequence of cuttingoff relations and as a consequence of connecting what had not before been connected. The empirical descriptions also showhow attempts to disentangle failed and how attempts to reduce complexities led to reactions and controversy.

9. Conclusions

In this study, we observe the problems of risk and coordination often discussed in the literature, as pointed out by Caglioand Ditillo (2008). However, guided by our theoretical framework, we also observe how many problems were related to theforming of actors, relations and products in relation to the interaction in the HOR. The study reveals how the HOR movedbetween the ultra-unstableness described by Thrane (2007) and the denser forms described in other studies (Mouritsenet al., 2001). In agreement with Mouritsen et al. (2001) and Kurunmaki and Miller (2011), we observe that mediatingtechniques, such as accounting and control practices (e.g. cost accounting and benchmarking) and organizationalarrangements (e.g. user forums) are essential since they pull entities together. Accounting and control contribute to thestabilization of HORs (Coad and Cullen, 2006; Kurunmaki and Miller, 2011; Mouritsen et al., 2001). However, while this, asindicated, is in line with observations in other studies, this study adds by also conceptualizing how accounting and controlnot only functioned as framing devices, but also became sources of overflows. What we observe is how accounting andcontrol contributed to the fragility of the HOR both through reconnecting the HOR to other complexities and through cuttingoff links in the process of disentanglement.

In line with Chua and Mahama (2007) we could observe how accounting and control caused controversies throughentangling the HOR to other complexities. For example, the cost allocation model reconnected ‘‘technical’’ issues about IT toissues about budget deficiencies in the welfare operations. What we also observe, however, is how the disentanglements thatthe framing process presupposes also led to overflows. When entities are pulled together, they are also cut loose. In order tocreate the ‘‘one shared system’’, the system had to be disentangled from the specifics of the operations performed in variouscontexts and various places. In line with Thrane and Hald (2006) we observe how the reduction of complexities in the HORcaused intra-organizational problems for the interacting parties. What we observe, in addition, is how these ‘‘externalities’’caused controversies and how this led to a questioning of the way the HOR was organized.

The theoretical point of departure, with the core concepts of framing and overflow, has supported the analysis of therelations between issues at stake in the organization of HORs and accounting and control initiatives. The study shows howaccounting and control helped resolve formative problems that we observed. Accounting and control both facilitated a denseHOR and the establishment of the HOR as an ‘‘enterprise’’ (Mouritsen and Thrane, 2006). But in addition, we add to theunderstanding by also conceptualizing how the very same accounting and control techniques facilitated overflows causingproblematizations of relationships, actors’ roles and definitions of products.

References

Akerstrom Andersen N. Public market – political firms. Acta Sociologica 2000;43(1):43–61.Barry A, Slater D. Introduction: the technological economy. Economy and Society 2002;31(2):175–85.Baxter J, Chua WF. Alternative management accounting research—whence and whither. Accounting Organizations and Society 2003;28(2–3):97–226.Beckert J. The social order of markets. Theory and Society 2009;38(3):245–69.


http://refhub.elsevier.com/S1045-2354(14)00013-6/sbref0005







G Model


Boedker C. Ostensive versus performative approaches for theorising accounting-strategy research. Accounting Auditing and Accountability Journal2010;23(5):565–95.

Briers M, Chua WF. The role of actor-networks and boundary objects in management accounting change: a field study of an implementation of activity-basedcosting. Accounting Organizations and Society 2001;26(3):237–69.

Broadbent J, Guthrie J. Public sector to public service: 20 years of contextual accounting research. Accounting Auditing and Accountability Journal2008;21(1):40–78.

Caglio A, Ditillo A. A review and discussion of management control in inter-firm relationships: achievements and future directions. Accounting Organizations andSociety 2008;33(7–8):865–98.

Callon M. An essay on framing and overflowing: economic externalities revisited by sociology. In: Callon M, editor. The laws of the market. Oxford: BlackwellPublishers; 1998.

Callon M. Introduction: the embeddedness of economic action. In: Callon M, editor. The laws of the market. Oxford: Blackwell Publishers; 1998.Callon M, Meadel C, Rabeharisoa V. The economy of qualities. Economy & Society 2002;31(2):194–215.Carlsson-Wall M, Kraus K, Lind J. The interdependencies of intra- and inter-organisational controls and work practices—the case of domestic care of the elderly.

Management Accounting Research 2011;22(4):313–29.Christensen T, Lægreid P. The whole-of-government approach to public sector reform. Public Administration Review 2007;67(6):1059–66.Christensen T, Lægreid P. The challenge of coordination in central government organizations: the Norwegian case. Public Organization Review 2008;8(2):97–116.Chua WF, Mahama H. The effect of network ties on accounting controls in a supply alliance: field study evidence. Contemporary Accounting Research

2007;24(1):47–78.Coad AF, Cullen J. Inter-organisational cost management: towards an evolutionary perspective. Management Accounting Research 2006;17(4):342–69.Cristofoli D, Ditillo A, Liguori M. Do environmental and task characteristics matter in the control of externalized local public services? Unveiling the relevance of

party characteristics and citizens offstage voice Accounting Auditing and Accountability Journal 2010;23(3):350–72.Das TK, Teng B-S. Between trust and control: developing confidence in partner cooperation in alliances. The Academy of Management Review 1998;23(3):412–91.Dechow N, Mouritsen J. Enterprise resource planning systems, management control and the quest for integration. Accounting Organization and Society

2005;30(7):691–732.Dekker H. Value chain analysis in interfirm relationships: a field study. Management Accounting Research 2003;14(1):1–23.Dekker H. Control of inter-organizational relationships: evidence on appropriation and co-ordination requirements. Accounting Organization and Society

2004;29(1):27–49.Dekker H, Anderson S. Accounting in networks—the transaction cost economic perspective. In: Hakansson H, Lind J, Kraus K, editors. Accounting in networks. New

York: Routledge; 2010.Fountain JE. Paradoxes of public sector customer service. Governance 2001;14(1):55–73.Goffman E. Frame analysis: an essay on the organization of experience. Boston: Northeastern Univ. Press; 1986.Granovetter M. Economic action and social structure. The problem of embeddedness. American Journal of Sociology 1985;91(3):48–510.Janssen M, Joha A. Emerging shared service organizations and the service oriented enterprise: critical management issues. Strategic Outsourcing An International

Journal 2008;1(1):35–46.Kastberg G, Siverbo S. The impossible split? A study of the creation of a market actor International Advanced in Economic Research 2008;14(1):65–75.Kurunmaki L, Miller P. Regulatory hybrids: partnerships, budgeting and modernising government. Management Accounting Research 2011;22(4):220–41.Lindvall J, Iveroth E. Creating a global network of shared service centres for accounting. Journal of Accounting and Organizational Change 2011;7(3):278–310.Llewelyn S. Managing the boundary. Accounting Auditing and Accountability Journal 1994;7(4):4–23.Lowe A. Accounting information systems as knowledge-objects: some effects of objectualization. Management Accounting Research 2001;12(1):75–100.Luhmann N. Social systems. Stanford, CA: Stanford University Press; 1995.McNulty Terry. Ferlie E. Process transformation: limitations to radical organizational change within public service organizations. Organization Studies

2004;25(8):1389–415.Meira J, Kartalis N, Tsamenyi M, Cullen J. Management controls and inter-firm relationships: a review. Journal of Accounting and Organizational Change

2010;6(1):149–69.Miller P, O’Leary T. Accounting and the construction of the governable person. Accounting Organizations and Society 1987;12(3):235–65.Mouritsen J, Justesen L. Effects of actor-network theory in accounting research. Accounting Auditing and Accountability Journal 2011;24(2):161.Mouritsen J, Thrane S. Accounting, network complementarities and the development of inter-organisational relations. Accounting Organizations and Society

2006;31(3):241–75.Mouritsen J, Hansen A, Hansen C. Inter-organizational controls and organizational competencies: episodes around target cost management/functional analysis

and open book accounting. Management Accounting Research 2001;12(2):221–5.Mouritsen J, Mahama H, Chua WF. Actor-network theory and the study of inter-organizational network-relations. In: Hakansson H, Lind J, Kraus K, editors.

Accounting in networks. New York: Routledge; 2010.Orlikowski W, Iacono S. Research commentary: desperately seeking the ‘‘IT’’ in IT research—a call to theorize the IT artefact. Information Systems Research

2001;12(2):121–2.Peters BG. Managing horizontal government: the politics of co-ordination. Public Administration 1998;76(3):295–312.Quattrone P, Hopper T. What is IT? SAP, accounting, and visibility in a multinational organisation Information and Organization 2006;16(3):212–50.Roberts J, Jones M. Accounting for self interest in the credit crisis. Accounting Organizations and Society 2009;34(6–7):856–67.Slater D. From calculation to alienation: disentangling economic abstractions. Economy & Society 2002;31(2):234–55.Strathern M. What is intellectual property after. In: Law J, Hassard J, editors. Actor-network theory and after. Oxford: Blackwell Publishing; 1999.Thrane S. The complexity of management accounting change: bifurcation and oscillation in schizophrenic inter-organisational systems. Management Accounting

Research 2007;18(2):248–72.Thrane S, Hald K. The emergence of boundaries and accounting in supply fields. The dynamics of integration and fragmentation. Management Accounting

Research 2006;17(2):288–309.Tomkins C. Interdependencies, trust and information in relationships, alliances and networks. Accounting Organization and Society 2001;26(2):161–89.Vosselman E, van der Meer-Kooistra J. Accounting for control and trust building in interfirm transactional relationships. Accounting Organizations and Society

2009;34(2):267–83.Wagenaar H. Governance, complexity, and democratic participation: how citizens and public officials harness the complexities of neighborhood decline. The

American Review of Public Administration 2007;37(1):17–50.






































































